Challenge 4th Generation Battle Facilities Discussion and Records

Hello, first time poster here. I've been getting back into the gen 4 frontier on emulator recently, and I wanted to start sharing some of my streaks.
Right now my best streak in the battle hall is 181 (on emulator) with garchomp using the following set:

Garchomp @ Focus Sash
Ability: Sand Veil
EVs: 4 HP / 252 Atk / 252 Spe
Jolly Nature
IVs: 30 HP / 28 Def / 1 SpA / 26 SpD / 30 Spe
- Earthquake
- Fire Fang
- Outrage
- Swords Dance

The set is basically the same as Nickm65's garchomp set, with the exception of some of the IVs, which aren't all perfect since I manually bred it and couldn't be bothered to spend more time trying for a better one. Here's a video of battles 171-182 for proof of the streak:


The streak ended due to an untimely freeze from ice punch lickylicky. I plan on doing at least a few more attempts with garchomp to try and get a streak above 200, but I'm interested in trying out a few different pokemon in the near future as well.
 
Time for some completely useless information that no one cares about, recently got an R4 so I decided to test an idea discussed in this post, and I am sad to say that you cannot cheese the Factory to get extra elevations or build artificially long streaks. Tested this with my Heart Gold, Platinum and R4 Soul Silver.

Started by clearing Round1 between my Heart Gold (Leader) & Platinum (Guest) with 7 swaps for Heart Gold, then I connected Heart Gold with R4 Soul Silver selecting Heart Gold as the Become Leader Option. What happened was that I got to play Round2 normally with the only weird things being that after clearing Round2 R4 Soul Silver received 10BP whereas Heart Gold received 11BP, and the Records Screen for R4 Soul Silver showed 7 wins, not 14.

Next I connected my Heart Gold (14 wins) with Platinum (7 wins) this time selecting Platinum as the Become Leader Option to see if maybe which Round is played is decided by the Leader, but again I proceeded to play Round3 for Heart Gold.

After that I started fresh and cleared Round1 again between my Heart Gold (Leader) & Platinum (Guest) with 7 swaps for Platinum this time, then I connected Platinum (Leader 7 wins as Guest) with R4 Soul Silver (Guest) hoping to get to play Round1 this time but again got to play Round2 normally.

The one thing I didn't test that could maybe get a different result is having 4 different Game Cartridges, let's say for example Heart Gold R4 Heart Gold R4 Soul Silver Platinum. You'd get Heart Gold to Round3 by doing Round1 linked with R4 Heart Gold, Round2 linked with R4 Soul Silver, then you'd connect R4 Soul Silver (7 wins but Round2 Clear) with Platinum (0 wins) and see if that would result in playing Round3 or Round2.

This is not something I'm interested to test out since while it would be funny if R4 Soul Silver (7 wins but Round2 Clear) got to play Round3 with Platinum (0 wins) it wouldn't really achieve anything in terms of getting extra elevations. Basically the only use I can think of if that is the case (if you get to play Round3 with 7 wins I mean) would be to give a friend's Game the ability to go straight to later Rounds without that being reflected in the Records Screen (provided you can clear the 1 Round)

So, to summarize, when connecting different games for Multis, it seems like the longest streak always gets prioritized. The possibility to give a different Game the ability to play later Rounds without that being reflected in the Records Screen exists, if someone feels like testing that I'd be interested to hear the results but I don't feel like playing for like 1.5 hours (in the best case scenario that I win 2 Rounds without losing on my first attempt) to clear 2 Rounds just to test that :)
 
Hello again, I have a few new streaks to report from this past week:

Battle Hall Singles: 186 with Garchomp (video proof)
Battle Hall Doubles: 270 with Metagross (vid1 vid2 vid3)
(both on emulator)
~~~
Details:
For Hall Singles, I did a few more attempts with the same Garchomp from my previous post, and ended up getting another run past Argenta:

Garchomp @ Focus Sash
Ability: Sand Veil
EVs: 4 HP / 252 Atk / 252 Spe
Jolly Nature
IVs: 30 HP / 28 Def / 1 SpA / 26 SpD / 30 Spe
- Earthquake
- Fire Fang
- Outrage
- Swords Dance

I didn't mention in my last post, but the order I've been using for my attempts is the same one that Nickm65 used for his 2nd place streak:

pre-170:
ICE, DARK, WATER, GHOST, GROUND, GRASS, FIGHTING, NORMAL, PSYCHIC, FLYING, BUG, DRAGON, STEEL, ROCK, POISON, FIRE, ELECTRIC
post-170:
ELECTRIC, FIRE, POISON, DRAGON, STEEL, ROCK, PSYCHIC, FLYING, BUG, GHOST, FIGHTING, NORMAL, GRASS, GROUND, DARK, WATER, ICE

The streak ended up dying to an unlucky freeze again, this time to regice. I'll probably do at least a few more attempts with Garchomp to try and get above 200.
~~~
For Hall Doubles, this is the team I used:

explosion (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 40 HP / 252 Atk / 4 Def / 212 Spe
Adamant Nature
- Explosion
- Iron Head
- Rock Slide
- Earthquake

cleanup (Metagross) @ Salac Berry
Ability: Clear Body
EVs: 36 Atk / 252 SpA / 220 Spe
Mild Nature
IVs: 30 Atk / 30 SpA / 30 Spe
- Endure
- Grass Knot
- Rock Slide
- Hidden Power [Fire]

(full disclosure, both of these mons were hacked in with pkhex)
The rationale behind using metagross is pretty simple: explosion is the strongest move in the game (effectively 500 base power), and metagross is the strongest explosion user, so it felt like a natural choice to try and make work. I looked at Chinese Dood and Grani's metagross sets, but I think they each have their own shortcomings (respectfully), so I tried to improve upon them as best I could.

For the explosion metagross: 212 Speed EVs hits 343 Speed with the choice scarf, which outspeeds starmie/raikou and everything slower. 252 Attack EVs and Adamant nature are used to maximize the damage output from explosion, and the remaining EVs are dumped into bulk. Explosion is obviously the main attacking move; it OHKO's the majority of opponents you face at the battle hall, with the exception of ghost types, bulky rock and steel types, slaking, regigigas, suicune, anything with the damp ability, and anything with a focus sash. The remaining moves on each metagross help deal with these exceptions. Iron head and earthquake help with rock types and steel types, respectively. Rock slide provides a spread move that doesn't hit the cleanup metagross, and helps deal with things like sash charizard.

For the cleanup metagross: Mild nature is used since the slightly reduced defense guarantees that explosion will activate the salac berry. 220 Speed EVs hits 344 Speed after the salac berry boost, which outspeeds the explosion metagross. 252 SpA EVs are used to maximize the damage output from grass knot and HP fire, and the remaining EVs are dumped into attack to increase the power of rock slide. Endure is used to ensure that the cleanup metagross doesn't die from a crit explosion or earthquake from its teammate. Grass knot helps with damp pokemon, HP fire helps hit things like scizor, bronzong, and skarmory, and rock slide is used to have another spread move.

I used the following order for battles 1-170:
GHOST, FIRE, GROUND, WATER, FIGHTING, STEEL, ROCK, DRAGON, PSYCHIC, NORMAL, GRASS, FLYING, DARK, ELECTRIC, BUG, POISON, ICE
The general rational behind this order is prioritizing types that metagross is weak to (fire/ground), types with damp (water/fighting/ground), and types that resist or are immune to explosion (ghost/steel/rock). After that I just did the types in order of what I perceived to be the most threatening. Post battle 170, I did the same order, but backwards to help maximize the length of the streak:
ICE, POISON, BUG, ELECTRIC, DARK, FLYING, GRASS, NORMAL, PSYCHIC, DRAGON, ROCK, STEEL, FIGHTING, WATER, GROUND, FIRE, GHOST
I don't know whether this is the "optimal" order to do for metagross, but it seemed to work well enough for me.

The streak ended up dying to lax incense magmortar, after I missed an earthquake on one of them.

I've been thinking about ways to improve the two sets for future runs, mainly in terms of the cleanup metagross's set: I think running either chilan berry, focus sash, or significantly more bulk on the cleanup metagross might allow me to forego running endure in favor of another attacking move. Thunder punch seems like it is probably better than grass knot at least after battle 170 since quagsire (the only damp mon immune to thunderpunch) doesn't appear post-170. Shadow ball could possibly replace rock slide to more reliably deal with ghost types. If an additional attack slot is freed up, I think either psychic or zen headbutt could be used to provide additional STAB coverage. I'm also open to ideas if anyone has any.
 
Last edited:
Very nice, I also really think adding the videos adds a lot (both in terms of transparency but also just interest).

As someone who has not touched hall in over a decade, i was curious if you actually have a threatlist for dual meta. Although magmortar "can" win i think EQ + endure is definitely the click in that matchup. After that a loss requires both a lax incense miss into fire blast hitting the EQ meta and/or for the endure meta to fail the 50% endure into flinch/crit and/or a further lax incense miss to happen. This isnt a dramatic improvement over just exploding but I think its worth double checking EQ rolls on less bulky opponents.

Thinking out loud: Entei seems like a larger problem (statistically). Heatran "can" win but is surely even less likely than mag
Ground doesn't look anywhere near as bad as i assumed before looking. Gastro and whiscash both have a similar matchup to Magmortar although they dont get guaranteed OHKOs with earth power i'm sure (which means the AI can behave a lot more randomly) at the cost of not having to rely on a 85% move.

Metagross shouldn't have issues with ghosts which is probably the biggest bonus for a boom strat.

I agree that it feels intuitively obvious that the streak could be improved by having an "A" and "B" team with slightly different responsibilities. There may be a few opponents where to be completely safe you need to just fire off EQ without endure (in which case sash or shuca works).

Also if it turns out that the only "real" threat is things that sometimes live boom, i wonder if it could be optimal to just run wide lens OR (dramatically cooler) to run dual scarf, and have the other metagross be scarf gravity (for the 5/3x universal accuracy boost) with enough bulk to always tank explosion or (perhaps necessary) to run this team at an xx2 or xx7 level (e.g. level 97) to assist on the metagross teammate survival since all moves are slightly weaker. This team would have to make a number of concessions though and is probably only sensible against types where scarf gravity + scarf explosion "always" works (plus the opponents arent fast enough to beat "slow" scarf meta).

Potentially even more insane if the above doesnt work; you could intentionally run self destruct and not explosion to help the teammate metagross survive. Again you'd have to carefully examine against which types this is viable. Even better; the metagross could run both explosion and self destruct, with self destruct for opponents where you dont want to endure.

Meta actually has quite a lot more interesting assist moves. Magnet rise probably doesnt work but icy wind is an exceptional move in doubles. Probably beyond the range of a sensible suggestion but metagross could also potentially run weather to assist on fire (rain dance) or for sash (sandstorm)

Regardless i think it would be cool to see you return with an A/B duo at some point (whatever the rationale is)!
 
Another week another streak:

Battle Hall Doubles - 321 with Metagross (proof: 171-220, 221-250, 251-300, and 301-322)
on emulator, as per usual

I put together a new set of teams this time:

Team A:

explode A (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 8 HP / 116 Atk / 172 SpD / 212 Spe
Adamant Nature
- Explosion
- Earthquake
- Rock Slide
- Thunder Punch

support A (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 160 HP / 72 Atk / 60 Def / 216 Spe
Impish Nature
- Gravity
- Rock Slide
- Protect
- Thunder Punch

~~~
Team B:

explode B (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 40 HP / 252 Atk / 4 Def / 212 Spe
Adamant Nature
- Explosion
- Earthquake
- Rock Slide
- Iron Head

support B (Metagross) @ Focus Sash
Ability: Clear Body
EVs: 56 HP / 252 SpA / 200 Spe
Modest Nature
IVs: 2 Atk / 30 SpA / 30 Spe
- Hidden Power [Fire]
- Grass Knot
- Psychic
- Protect

(all obtained with pkhex btw)
Team A handles ice, psychic, flying, water, and fire. Team B handles bug, grass, rock, steel, and ground. The other types are shared between the two teams. This was the order I did the types after battle 170:
171-180 (B)PoisonDragonNormalDarkFightingBugGrassRockSteelGround
181-190 (A)ElectricGhostIcePsychicFlyingWaterFirePoisonDragonIce
191-200 (B)NormalDarkFightingElectricGhostBugGrassRockSteelGround
201-210 (A)PsychicFlyingWaterFirePoisonDragonIcePsychicFlyingWater
211-220 (B)NormalDarkFightingElectricGhostBugGrassRockSteelGround
221-230 (A)FirePoisonDragonNormalDarkIcePsychicFlyingWaterFire
231-240 (B)FightingElectricGhostBugGrassRockSteelGroundPoisonDragon
241-250 (A)NormalDarkFightingElectricGhostIcePsychicFlyingWaterFire
251-260 (B)BugGrassRockSteelGroundPoisonDragonBugGrassRock
261-270 (A)NormalDarkFightingElectricGhostIcePsychicFlyingWaterFire
271-280 (B)SteelGroundPoisonDragonNormalBugGrassRockSteelGround
281-290 (A)DarkFightingElectricGhostIcePsychicFlyingWaterFirePoison
291-300 (B)DragonNormalDarkFightingElectricBugGrassRockSteelGround
301-310 (A)GhostIcePsychicFlyingWaterFirePoisonDragonIcePsychic
311-320 (B)NormalDarkFightingElectricGhostBugGrassRockSteelGround
321-330 (A)FlyingWater*FirePoisonDragonIcePsychicFlyingWaterFire
331-340 (B)NormalDarkFightingElectricGhostBugGrassRockSteelGround
*I lost on this battle
~~~
Spread details:
For team A, the explosion metagross is EVed to outspeed starmie/raikou and survive one eruption from QC entei. The moves are the same as in my previous team with the exception of thunder punch over iron head, which helps with water types. The support metagross is EVed to outspeed its partner and survive a crit explosion without the need for endure/protect/sash.
scarf gravity
The idea of running scarf gravity was really appealing to me especially after losing my last streak to lax incense magmortar, so I decided to run with it. Gravity barely gets used with this team, but it's still really nice to have since it guarantees the win against opponents which would otherwise be very annoying to deal with (magmortar, starmie, jynx, piloswine, etc.). Protect is used specifically for QC entei, so that the support metagross can guaranteed stay alive after eruption+explosion. Rock slide is mainly for sash charizard, but it's also just generally good having a spread move on the support metagross. Thunder punch helps with water types.

For team B, the explosion metagross is identical to the one used in my previous streak. The support metagross runs sash to live explosion in cases where I don't want to protect. HP fire and grass knot help to reliably deal with mons like scizor and donphan, respectively. Psychic notably ohko's poliwrath and otherwise acts as a strong stab move to use when HP fire/grass knot don't hit super effectively.
~~~
Threats:
Right now, knowing what I know, I consider the biggest threats to this team to be (not ordered):
-charizard (have to rely on rock slide's accuracy)
-infernape (fake out+flare blitz)
-entei (QC eruption)
-poliwrath (hypnosis+can't reliably ohko with team A)
-omastar (can't reliably ohko with team A)
-suicune (reflect+explosion doesn't always ohko)
-rotom (can't reliably ohko)

There are plenty of other mons that CAN beat these teams with enough bad luck (heatran, dusknoir, alakazam, bastiodon, manectric, gengar, tyranitar, the list goes on), but I think the ones above are by far the worst to deal with (though there may be some I've missed).
~~~
How I lost:
I ran into suicune on battle 322.
Turn 1: I opted to go for explosion+thunder punch to get a guaranteed kill on at least one suicune. The other suicune lives and goes for reflect. It looks like it's around 33% health after the sitrus berry, so thunder punch should still kill in 2 hits even with low rolls ( 72 Atk Lv. 100 Metagross Thunder Punch vs. 0 HP / 252+ Def Lv. 96 Suicune through Doubles Reflect: 68-82 (21.1 - 25.5%) ).
Turn 2: I thunder punch. Suicune lives. Suicune goes for hydro pump, it hits and I live.
Turn 3: I thunder punch. Suicune STILL lives (???), but gets paralyzed. Suicune doesn't get fully para'd and hits hydro pump and I lose.

I'm still completely lost on how suicune didn't die to explosion+two thunder punches even after reflect+sitrus. Even with the lowest possible rolls on every single one of my attacks it still should've killed unless I calc'ed it wrong. idk
~~~
Anyways, I still think these teams have potential to go further, so I'll be doing more attempts with them to try and beat Jumpman and TRE's Hall Doubles records.
 
Last edited:
Threats:
-charizard (have to rely on rock slide's accuracy)
-infernape (fake out+flare blitz)

Congrats on the streak, just my two cents on zard and ape being on your threat list. I don't really see why you consider them a problem.

-Charizard will never Blast Burn from full so you don't need to click Rock Slide and can just double tpunch instead, targeting both of them turn1 to avoid the Protect WoW combo and then kill the one tpunch connected with or just kill both turn2 in case none of them protected.

-Infernape sees a kill with Flare Blitz so it should never Fake Out meaning you have a free Explosion win.

Just keep in mind that I'm speaking from experience which is almost exclusively from Hall Singles and pre170 so I could very well be wrong, maybe the scenarios you're scared of have happened to you and you're right to be worried, just read your report and I found it weird that you had these 2 down as threats is all.
 
-Infernape sees a kill with Flare Blitz so it should never Fake Out meaning you have a free Explosion win.
good point, I agree with this; just didn't occur to me in the moment.

-Charizard will never Blast Burn from full
I didn't know this was the case. Is the AI hard-programmed to never use blast burn at full health?

I'm still completely lost on how suicune didn't die to explosion+two thunder punches
Turns out I did miscalc. I wasn't aware of how reflect works in doubles, Tpunch does less when reflect is active and only one opponent is alive. Now that I know that, I think explosion+rock slide is probably the better play vs suicune.
 
I didn't know this was the case. Is the AI hard-programmed to never use blast burn at full health?

Personally I've never seen Blast Burn/Frenzy Plant/Hydro Cannon used from full and I've based Hall strategies around that knowledge and never had a problem, and from facing off against those moves in Factory I know that they're pretty much used only when on low health, you could ask in the Discord for more info, someone more knowledgeable than me could maybe give exact percentages from the decomp, potential differences between Singles/Doubles etc. etc.
 
Tsito is correct, the "Recharge turn" logic is as follows:
If the opponent (i.e. the player) would resist or is immune to the move, score -1.

If the attacker's (i.e. the AI attacking) ability is Truant, 68.75% chance of score +1.

If the attacker is slower than the opponent and the attacker's HP is >= 60%, score -1.

If the attacker is faster than the opponent and the attacker's HP is > 40%, score -1.
A move scoring -1 under regular circumstances means it cant be selected.

For context, the strongest "normal" move will always be score 0 unless it is in some way impossible or useless (examples; dream eater vs awake target, EQ vs immune pokemon, etc) Most status moves that work/ do not have a significant drawback get a score of 0 or higher.


Still a great streak
 
HGSS Factory Lvl50 Singles: 156 (Emulator)

Warning this write up is 30k words long but does have video links for the entire run and the later rounds (14 onwards) include (patchy) live commentary.

In another post I will do a guide, tierlist of all 950 sets, among other general thoughts on the factory.

While the full run was livestreamed on twitch and has been copied over to youtube, I'll go through a rundown anyway in case i never get around to doing a proper edited video with post-commentary explaining thought processes. This runthrough is necessary as it highlights misplays.
[If I ever do get around to editing ~20 hours of livestream footage down to a more manageable video, it will go here]

Until then, I'll link timestamped youtube videos of each round and add some thoughts. This is both a writeup and a kinda "guide" because lets be real, most guides are shit without actual examples.
I encourage any serious player or factory player or whatever to watch/ read along and really thoroughly think about what you would do before I provide any comments. This probably gives you some insight into the depth involved in the factory. Frontier players should already know this; how deep a single team of 3 can go. Well "semi" randomised teams of 3 is a bit more complicated to constantly be on top of (and perhaps always makes the player look like an idiot? Idk)

I also know there is a long standing tradition of only doing write-ups starting at 28 wins. This made sense back in the era where no one recorded anything but I see no reason not to start at round 1, especially if this is also meant to be a "guide"

To understand the formatting: I refer to sets by their standard index, this is the name you'll find on resources on smogon as well as other websites. For example, the first Abomasnow set listed in the data is "Abomasnow1".
I will list my team before each match, as well as the opponents full team.
Occasionally, the opponent's set may be unknown. If so I'll leave it as "?"
(44.5 mins if the timestamp link fails)

(Fully random AI, draft is all 0IV, opponents 1-6 are 0IV, opponent #7 will be 4IV + tier 2 pokemon)

Swapping every single round here is important for elevations in later rounds and costs almost nothing.

Round 1 Lvl50 is free but reckless choices do make it possible to lose. My main objective in round 1 is to simply take the obviously strong pokemon that wont randomly get OHKO'd and this is arguably more important than taking the best sets.

Draft: :Kabuto: / :Snorunt:/ :Spinda: / :Chatot: / :Shuppet: / :Smoochum:

My rough tiers for these factoring in wanting games to be over in under 8 turns ideally:
:Kabuto: C
:Snorunt: D
:Spinda: S
:Chatot: B
:Shuppet: D/E
:Smoochum: B/C

:Spinda: /:Chatot: /:Smoochum: is therefore the clear pick and this team is already like 99.9% to 7-0 but I will probably be forced to make it worse via swapping.

Battle 1: :Anorith: / :Bronzor: / :Wailmer:
:Bronzor: is easily an S tier round 1 pokemon and getting rid of :smoochum: and its base 45/15 physical defense basically secures the rest of the round.

Team: :Spinda: /:Chatot: /:Bronzor:

Battle 2: :Sableye: / :Graveler: / :Shellos:
Picking up :Graveler: as a lead mainly because of its ability to win battles faster

Team: :Graveler: /:Chatot: /:Bronzor:

Battle 3: :Wailmer: / :Skiploom: / : :Glameow: :
Due to the forced skip I take :Glameow: over :Chatot: . This is a downgrade but 7/7 swaps is worth it.

Team: :Graveler: / :Glameow: /:Bronzor:

Battle 4: :Plusle:/ :Kricketot:/ :Swablu:
:Plusle: is one of the best pokemon in round 1 as a compromise between fast battles and safety. :Glameow: leaves here

Team: :Graveler: /:Plusle:/:Bronzor:

Battle 5: :Dustox:/ :Chatot: / :Smoochum:
This is a reasonably scary team if random AI gets lucky and if my team wasnt already good after 4 swap options.
:Dustox: is another incredible mon for early rounds and like :graveler: its quite good at cheesing the tier 2 pokemon that will be in battle 7.
I drop :bronzor: here but in terms of ruthless winrate it would be better to drop :plusle:.

Team: :Graveler: /:Plusle:/:Dustox:

Battle 6: :machop:/ :baltoy:/ :piplup:
Very weak team.
This is annoying because a swap is forced and battle 7 is the only realistic loss point.
Luckily the Battle 7 preview reveals a joke pokemon (zangoose1) and so i can simply swap for the best lead matchup as :plusle: will beat kingler1 despite its wacan.

Team: :Machop:/:Plusle:/:Dustox:

Battle 7: :Sealeo:/ :Kingler:/ :Zangoose: (4IV/ Tier 2)
For a tier 2 team this is pretty weak.
(53 mins if the timestamp link fails)

(Fully random AI, opponents can be 0IV/ tier 1 OR 4IV/ tier 2, Battle #7 is against tier 3, the draft is 5 Tier 2 and 1 Tier 3 Elevation)

This is the final round where I think it's correct to just always swap no matter what to get 14/14 going into round 3. Older wisdom is to go for 28/28 swaps going into round 5. I say that it isn't worth tanking winrate to get a single additional elevation and I have enough respect for the semi-random AI.

The final Battle of round 2 is against a tier 3 trainer. The jump from tier 2 -> tier 3 pokemon is quite large, and so this match is losable to a few unlucky (for the player POV) enemy AI decisions in that fight.

What is annoying is if Battle 6 is against a tier 1 trainer. This usually leaves the team with 1 bad pokemon. The aim is therefore to find 2 good pokemon.

Draft: (Tier 2) :Tangela:/ ::Banette::/ :Golbat:/ :Electabuzz:/ :Monferno:
(Tier 3) :Luxio: (intimidate)

:Tangela: is pretty good
::Banette:: isn't
:Luxio: is a pretty good elevation. (Elevations are rarely turned down since tier 3 is much better as tier 2 as mentioned above)
:Monferno: is ok
:electabuzz: is good
:Golbat: is good
Common sense usually prevails against random AI, and common sense says to not take 3 ground weak pokemon

The ruthlessly "best team" is probably :Luxio:/:Tangela:/:Golbat: but speed somewhat matters and the below take is faster while still being very likely to beat random AI trainers.

Team: :Monferno:/ :Luxio:2/ :Golbat:1

Battle 1: (Tier 1) :Elekid:/:spoink:/:roselia:
Once again this is a semi awkward position. :Elekid: is by far the best swap option. I do not want to drop :golbat: and :luxio: is the elevation/ carry.
So as dumb as it may look to someone not familiar with factory swap elevations, the correct decision here is definitely :monferno: >< :Elekid:

Team: :Elekid:1/ :Luxio:2/ :Golbat:1

Battle 2: (Tier 2) :Magmar:/:Hitmonlee:/:Carnivine:
:Magmar: >< :Elekid: and almost a return to the original draft

Team: :Magmar:1/ :Luxio:2/ :Golbat:1

Battle 3: (Tier 1) :Nidorina:/ :Snover:/ :Bulbasaur:
:Nidorina: is the only real option and I decide to drop :Golbat: instead of :Magmar:

Team: :Magmar:1/ :Luxio:2/ :Nidorina:1

Battle 4: (Tier 2) :Bibarel:/ ::Jumpluff:: / :Charmeleon:
:Bibarel: is strong by tier 2 standards and a lot better than :nidorina:.

Team: :Magmar:1/ :Luxio:2/ :Bibarel:1

Battle 5: (tier 2) :Quilava:/ :Monferno:/ :Golbat:
:Bibarel: >< :Quilava: as the flex spot and after seeing cherrim in the forecast.

Team: :Magmar:1/ :Luxio:2/ :Quilava:1

Battle 6: (tier 2) ::Banette::/ :Cherrim:/ :Hitmontop:
I would not usually take :cherrim:, but the forecast shows :Omastar: and starting the battle 3-2 ahead sounds good

Team: :Cherrim:1/ :Luxio:2/ :Quilava:1

Battle 7: (Tier 3) :Omastar:/ :Pidgeot:/ :Kingler:
I will always respect this battle's ability to beat me if i dont pay attention
This time around, random AI is true to its name and doesn't get past :Cherrim:
(1 hour 3 mins if the timestamp link fails)

("Basic" AI, Tier 2 and 3 opponents (tier 2 now has access to Tier 4 pokemon but at 4IV), Thorton (12IV + Tier 4) battle 7, draft is 4 x tier 3 and 2x tier 4 )

(For those wondering what Basic AI is - it essentially avoids doing obviously bad moves. EQ on levitate, Fake out on the second turn, etc)

Ideally round 3 is another round where the player swaps every single time

I do think that if the team is bad and the swap options approaching thorton1 suck, that it is reasonable to not swap. A single extra round 5 elevation isnt worth tanking the winrate against thorton1 by perhaps 20-40% (which is what a bad tier 2 pokemon may do)

While the jump in power from tier 3-4 is quite big, honestly the jump from 2-3 is even larger. Swapping for a tier 2 mon really sucks.

Draft: (Tier 3) :Jumpluff:/ :Grovyle:/ :Bayleef:/ :Bellossom:
(Tier 4) :Exploud:/ :Venusaur:

5 grass pokemon feels like a curse. Basic AI is very capable of winning the next fight. The next match preview showed :Swalot:1.

At the same time, this is what I enjoy about factory. I would normally never pick some of these but I'm going to be forced into at least 2.

:Bayleef: and :Grovyle: immediately disqualify themselves by being weak both offensively and defensively
:Exploud: and :Venusaur: are somewhat forced here as below average elevations that are still above average compared to Tier 3s.
:Bellossom: vs :Jumpluff: is closer than it may seem. While :Bellossom: is a better set it would be a nightmare to run into a fire type after I set sun for it.
The :exploud: lead is somewhat forced as i suspect Swalot's 3 stockpile spit up crit can OHKO :venusaur: and i see no reason to risk such things.

Team: :Exploud:1 / :Venusaur:1 / :Jumpluff:2

Battle 1: (Tier 2) :Swalot: / :Magneton: / :Wigglytuff:
:Magneton: is the only sensible swap option. :Jumpluff: isnt good.

Team: :Exploud:1/ :Venusaur:1/ :Magneton:1

Battle 2: (Tier 2) :Machoke: / :Quilava:/ :misdreavus:
I trust :machoke: over :quilava: for the same reason i trust fighting types in general a lot more than fire types

Team: :Exploud:1 / :Venusaur:1/ :Machoke:1

Battle 3: (tier 2) :Piloswine: / :hitmonchan: / :relicanth:
Impressively water pulse was a guaranteed 2hko on pilo, hyper voice not required. This is roughly what +8 IVs means for those trying to get a feel for it.
There are once again no good swap options here. Tier 2 trainers are annoying. Thorton is far enough away that i'll continue swapping.

Team: :Exploud:1/ :Venusaur:1/ :Relicanth:1

Battle 4: (Tier 2) :Persian:/ :misdreavus:/ :arbok:
If there is something that's a tiebreaker between terrible pokemon its generalised utility, which :misdreavus: has. Leaving :arbok: behind to not take 2 poisons.

Team: :Exploud:1/ :Venusaur:1/ :Misdreavus:1

Battle 5: (Tier 3) :Magmar:/ :Vigoroth:/ :Banette:
Tier 3 battles are harder to win but a lot better as giving realistic swap options for the incoming Thorton on Battle 7.
:Vigoroth:'s endure set is a pretty strong gimmick by frontier standards when facing non-random AI (thorton).

Team: :Exploud:1/ :Venusaur:1/ :Vigoroth:2

Battle 6: (Tier 2) :Wormadam:/ :Sudowoodo:/ :Carnivine:
It would have been completely acceptable to not swap here in certain circumstances. R4 is still basic AI, and the elevations for R5 are somewhat overrated. R6 and 7 elevations are good but its "just 1" that is missed out on.
However, the preview shows :Togekiss:1. This is a set that sudowoodo can beat, even if its hustle. Meanwhile :Exploud: can easily lose to it even with chople.

Team: :Sudowoodo:1/ :Venusaur::1/ :Vigoroth:2

Battle 7 (Frontier brain - 12iv) (Tier 4) :Togekiss:/ :Lucario: / :Tentacruel:
Sudo misses the rock slide roll on :togekiss: but Faint attack is guaranteed to KO through the healing.
:Lucario:1 is extremely dangerous but as with most things in the frontier, danger breeds predictability. :Lucario: will always aura sphere :vigoroth: and set up 1hp reversal.
I prefer this significantly to sending in :venusaur: and amnesia. The way that flash cannon would get multiple chances to crit or spdef drop (as well as Leaf storm's 81% acc and having to hit that twice) make it a lot less comfortable than it may seem at first.
I would certainly estimate the total chance of :venusaur: being hax'd as greater than 10% - which is the chance 1hp reversal misses.
With :tentacruel:1 in last it is likely this small decision did not matter. Still, its better to use resisted 100% acc reversal even in a game that looks like its already won.
(1 hour 15 mins if the timestamp link fails)

("Basic" AI, Tier 3-4 Opponents + Tier 5 on battle 7, draft is 3x Tier 4 and 3x Tier 5)

Round 4 is usually easier than Round 3, although its always worth respecting the tier 4 opponents specifically as pokemon like :Infernape:1 or :Garchomp:1 really only have to get a few "lucky" clicks in to win.
At the very least the Tier 2 trash is gone, from here on you can largely assume everything has a real set or what GF must have thought was a real set.

The typical wisdom is to once again go for full swaps here. Unlike the Thorton round I agree with this, and since i'm at 21 I might as well go for 28. If i was at 19 or 20, i'd probably not bother going much beyond 21.

Draft: (Tier 4) :Altaria:/ :Absol:/ :Probopass:
(Tier 5) :Abomasnow: / :Steelix:/ :Cradily:

This is another bad draft.
Usually you want the Tier 4s to carry but :Absol:/ :Probopass: are almost unpickable.
:Cradily:2 is several tiers below them, as its ability to do anything is often severely limited.
:Altaria: is ok in the context of the enemy AI still being random.
:Steelix:'s fang set is surprisingly good, particularly pre-49 where I will know if a set is physical/special/mixed (tl;dr is if its physical then :steelix: can take it on most likely).
:Abomasnow: (and leech sets in general) are quite good pre-28. They're worse than real sets of course but if you squint the random AI kind-of has truant and you mostly just dont want to accidentally get hax'd or pick a pokemon thats too frail.

:Altaria:1/ :Abomasnow:2/ :Steelix:2 - is something of a joke team and only kind of safe because of Basic AI. In a later round, I would probably be forced to go :Absol:1(!) / :Steelix:2 / :Altaria:1

Battle 1: (Tier 3) :Qwilfish: / :Dodrio: / :Kingler:
Against smart AI i would have been forced to play this almost entirely different, but going :Steelix: is actually a risk against basic AI which could easily aqua tail turn 1.
Despite being Tier 3, :Dodrio: is a massive upgrade over :Abomasnow:. Doubling up on flying types is totally fine, and a lot safer than relying on a bad set to do anything. Dead things are less scary than seeded things.

Team: :Altaria:1/ :Dodrio:2/ :Steelix:2

Battle 2: (Tier 4) :Granbull: / :Flygon: / :Crobat:
Rare example of a resistance switch from the opponent. It's easy to see how if random ai gets a few lucky turns in how quickly you can lose. This is why the early rounds of lvl50 are easier than open level.
Still, in classic factory fashion what was a scary matchup turns into a blessing the second you win. :Flygon:1 is an incredible pokemon

Team: :Flygon:1/ :Dodrio:2/ :Steelix:2

Battle 3: (Tier 4) :Swalot: / :bellossom: / :qwilfish:
All 3 of these are ok takes. :Bellossom: stands out as the simplest take. Although I'm now triple weak to ice, its very unlikely random AI would punish this.

Team: :Flygon:1/ :Dodrio:2/ :Bellossom:2

Battle 4: (Tier 3) :Machoke: / :Pelipper: / :Raticate:
Playing around counter on turn 1 was a mistake. While Smart AI is almost guaranteed to counter in this scenario, with basic AI is pretty much a 1/4. The good news is that after sand, :dodrio: is guaranteed to KO.
However 255 def base 100 :pelipper: is something that enough people probably arent familiar with. Surf or Ice Beam could have led to an embarrassing loss although with :raticate: last, :bellossom: would have won the 1v2. :Bellossom:2 is a surprisingly fine pokemon.

Its important to underline that if i lost here after the first turn crit - it was user error and not just "hax". There was an objectively better decision to be made and the factory almost slapped me for not doing it.

Team: :Flygon:1/ :Dodrio:2/ :Pelipper:2

Battle 5: (Tier 4) :Kingdra: / :Sceptile:/ :Tyranitar:
Not sure there is necessarily a correct swap between :kingdra: and :Tyranitar:. Against smart AI its :kingdra:. Against dumb AI its probably :Tyranitar:.

Team: :Flygon:1/ :dodrio:2/ :Tyranitar:1

Battle 6: (tier 4) :leafeon:/ :arcanine: / :crobat:
I would not normally take flash fire :arcanine: but its probably the best option and Tier 5 pokemon (set 2) are often more miss than hit

Team: :Flygon:1 / :Arcanine:1 / :Tyranitar:1

Battle 7: (Tier 5) :Ninetales:/ :Rampardos:/ :Jynx:
A simple win for :Flygon:
(1 hour 28 mins if the timestamp link fails)

(Smart AI, Tier 4-5 opponents + Tier 6 on battle 7, draft is 4x tier 6 and 2x tier 5)

Round 5 is, in my opinion, the hardest round of the factory.
The reason for this is that Tier 4 opponents specifically are deadly, and the players' draft options cannot be tier 4 or 7 but instead tier 5 or 6 only. This is never guaranteed to happen to the player again (although it is possible after Thorton 2).
(Tier 5 and 6 are often either meme sets, focusing on the weaker attack stat (e.g. physical :Manectric: or special :medicham:). Some tier 6 sets are very good but the variance is high)
After a few swaps you usually enter more safe territory.
It's important to double underline the "Smart" AI in effect. This isn't necessarily a bad thing as what it actually means is the AI is much more predictable.
However, the smart AI is definitely capable of closing out simple wins, and the main job as the player with a lot more freedom is therefore to ever deny the AI an opportunity to easily win.

Draft: (Tier 5) :Lickilicky:/ :Houndoom:
(Tier 6) :Golem:/ :Abomasnow:/ :Nidoqueen:/ :Donphan:

By Round 5 standards, this is actually not a terrible draft.
:Lickilicky:2 and :houndoom:2 are ok although I wouldn't really be happy with either on a team.
:Golem:3 does at least have some real moves, although it comes with :golem:'s horrendous typing and terrible speed. This almost always disqualifies :golem: from drafts unless we're talking about truly cursed drafts.
:Donphan: is basically a big EQ button which - while it can work - is inflexible.
:Nidoqueen: and :Abomasnow: are serviceable though as real pokemon with real attacks.

There are a lot of teams that could be made with these 6 and I wouldn't necessarily call them a mistake although I would question a few. So long as :Golem: is left out a lot of things are logical.

:Houndoom: forces its way onto the team simply by virtue of being fast. I cannot understate how important it is to have the option for this, its one of the main sources of active counterplay the player has.
If it hadnt already made the team, :houndoom: would also have forced its way on after the ice speciality was revealed in the battle 1 preview.
:Abomasnow: and :Nidoqueen: join as things that I mostly trust and things that seem reasonable into an ice trainer.

Team: :Houndoom:2/ :Abomasnow:3/ :Nidoqueen:3

The team is bad overall but probably slightly above average by Round 5 standards.

Battle 1: (Tier 5) :Flareon:/ :Dewgong: / :Glaceon:
This was a very uncomfortable battle. :Flareon: wasnt possible to safely switch out from.
The obvious switch on :dewgong: is :Abomasnow: but this potentially neuters :houndoom:'s ability to do something when it returns.
Endure is not the safe play it may appear against :dewgong: because both signal beam and ice beam have 10% secondary effects (fire fang is not an active thaw move).
After :houndoom: missed so many thunder fangs so as to actually faint the possibility of losing went up.
:Glaceon: last comes close to a punish. Hitting focus blast in snow cloak is uncomfortable, but it would have healed out of blizzard 2hko range with leftovers.

With hindsight, it's probably optimal to just go :nidoqueen: on :Flareon:. While :nidoqueen: does a lot against most ice pokemon despite its type disadvantage, it is too heavily punished by others.
:Houndoom: + :Abomasnow: is not a very safe final duo and :Houndoom: first is almost forced.
I should have correctly identified that the chance of :houndoom: missing a 5% thunder fang and then 3/7 25% thunder fangs was somewhat high (in comparison to the hax required for :nidoqueen: to lose)
While :houndoom: - :flareon: feels "very safe" (The AI has a 50% chance to guess wrong on flash fire for :houndoom: each turn), its not that hindsight-y to say that it was shaky at best.

Now :houndoom: very much loses its placeholder status as weak and a bit of an anti-ice gimmick.
of the swap options, :Dewgong: does the most defensively as I really hate being weak to water. :Glaceon: is slightly stronger but much more one dimensional set.
With :nidoqueen: and :Abomasnow: I am not worried about either electric or rock coverage causing serious problems for lead :dewgong:. The team still needs a real lead and/or something fast.

Team: :Dewgong:2/ :Abomasnow:3/ :Nidoqueen:3

Battle 2: (Tier 4) :Garchomp: / :ampharos:/ :yanmega:
:Garchomp:1 is usually scary but :abomasnow: being on the team with perfect accuracy Blizzard means for once its not a problem.
The last 2 are not threats to :Nidoqueen:.

:Garchomp:1 is one of the most obvious takes in factory, particularly in a round where the draft locks the player out of set1 and set4 pokemon.
It's definitely best positioned here as solving the lead problem, although :nidoqueen:/ :abomasnow: is a weaker backline for supporting it.

Team: :Garchomp:1 / :Abomasnow:3/ :Nidoqueen:3

Battle 3: (Tier 5) :Tauros:/ :Donphan:/ :Electivire:
set 2 trainers are rarely and although this one has above average pokemon for set2, :nidoqueen: and :garchomp: are both rip through here.

:Tauros: over :Abomasnow: is a swap i make for extra safety. :Garchomp: and :nidoqueen: should cover the majority of games, and its simply safer to not have a pokemon in the back with a 4x weakness
when its partner cannot easily defensively offer pivot or switch in support. :Tauros:' speed is also a much more welcome assist on anything that the other 2 may struggle to finish off.
On a weaker team, :Abomasnow: would probably be kept over :Tauros:.

Team: :Garchomp:1 / :Tauros:2 / :Nidoqueen:3

Battle 4: (TIer 4) :Poliwrath: / :porygon-Z: / :Armaldo:
This battle is easy but highlights an example of small optimisation.
If 2 moves can KO the opponent, its best to use a different one in case the previously used move causes an immunity or resistance switch.
Adaptability :Porygon-Z: is - in most scenarios - a straight up upgrade over :Nidoqueen: and another example of a safety swap. A double normal bench is not even e.g. :Lucario: weak as :tauros: is faster.

Team: :Garchomp:1 / :Tauros:2 / :Porygon-Z:1

Battle 5: (Tier 4) :Flareon:/ :Ludicolo:/ :dusknoir:
As this is another easy win, I'll again highlight how decomp knowledge assists here:
:Ludicolo: sent in second against :Garchomp: guarantees that the 3rd pokemon isn't ice or dragon type and it also guarantees no ice move and makes a dragon move very unlikely - although something like twister would be possible if a set that had it was a valid option.
In such a scenario, often it would be logical to save :Garchomp: for the 3rd pokemon.
But SD :Ludicolo: is somewhat annoying to give a free turn to, :Garchomp: 2hkos it anyway, and its got a 75% chance of using Swords Dance here as it will definitely not see a kill on :Garchomp:.

As the preview shows a water specialist, I strongly consider taking :Ludicolo: and lookup exactly what runs ice coverage for :Garchomp: and what the likely plan is against such pokemon.
It would be better for battles 6 and 7 to not take :Ludicolo: and the major risk (That something is frozen/ crit/ etc as :Garchomp: is switched out or after it faints) does not meaningfully change with the switch.

Team: :Garchomp:1 / :Tauros:2 / :Porygon-Z:1

Battle 6: (Tier 5) :Dragonite: / :Floatzel: / :Tentacruel:
While :dragonite:2 prefer ice punch due to the slight increase in score the AI gives 4x, Switching is not entirely safe as dragon rush can be selected and would OHKO via a crit.
I decide that I would rather keep :Garchomp: alive than :porygon-Z:, and so sack :Porygon-Z: for the chip damage that dragon claw needs.
:Floatzel: coming in second is nice as it guarantees no ice coverage in 3rd and because :Floatzel: is frail enough to drop to a single EQ.

Dark is somewhat rare in preview. :Weavile:3 isn't actually a problem (no ice moves). :Tyranitar:3 can actually survive :tauros: EQ + :Porygon-Z: Adaptability Tri attack.
Taking :Dragonite:2 is mostly about also giving freedom to use :Garchomp:/ let it faint in the early matchups and know I'm not locking into a losing endgame vs :Tyranitar:.

Team: :Garchomp:1 / :Tauros:2 / :Dragonite:2

Battle 7: (Tier 6) :Shiftry: / :Lickilicky: / :Houndoom:
Although :Garchomp: can beat :shiftry:, Leaf storm t1 is guaranteed by smart AI and :Dragonite: can easily take it. :Dragonite: is less valuable.
:Lickilicky:3 being revealed is a problem. The AI is very likely to curse into rest on turns 1 and 2, as both of these moves are very likely to score considerably above 0.
The plan is definitely to EQ in order to force rest as EQ will do more than 40% and cause rest to have a ~97% chance to be selected.
On the rest turn, :Garchomp: can sandstorm which will push :Lickilicky: a lot closer to double rest range after it gets a single turn (which would likely be curse #2).
This should ensure the best chance of avoiding double return and will give :Garchomp: and :Tauros: an additional 2 turns to crit after the curse boosts.
Instead of all of this, EQ simply crits immediately, meaning :Lickilicky: as well as the chance to ever lose drop. (What set3 dark pokemon could 2-0 :Garchomp: and :Tauros:?)

With the way this played out, perhaps the :dragonite: swap (as well as the decision to let :dragonite: take leaf storm) could be reviewed.
Mixed team coverage is largely overrated in comparison to just taking strong sets, however :Porygon-Z: >< :dragonite: was a close call regardless and again if i lost here it would have been my fault for dropping Porygon.
As-is, the way the :lickilicky: plan turned out reveals how important it is to understand the AI's decision making. Anything except attacking the first turn :lickilicky: is out would have been a serious mistake.
(1 hour 52 mins if the timestamp link fails)

(Smart AI, Tier 5-6 opponents + Tier 7 on battle 7, draft is 4x tier 7 and 2x tier 6)

Round 6 is usually the easiest round where smart AI is involved.
The player gets access to multiple Tier 7 (Set 4 / 24IV) pokemon which are usually the best sets for that species in the draft.
Meanwhile the enemies are all stuck on tier 5 and 6 which are largely suboptimal/ gimmick/ joke/ meme/ wacky sets. Tier 6 is definitely the greater threat but also the best chance to upgrade swap.
However battle 42 specifically (also Tier 7) deserves some respect and is one of the deadlier battles pre-49.

Draft: (Tier 6) :Steelix:/ :Umbreon:
(Tier 7) :Gengar:/ :Exploud:/ :Skuntank:/ :Donphan:

:Umbreon:3 is up there with the worst pokemon in the factory. Some gimmicks are worth risking a run on (e.g. Hypnosis), i dont think I would ever want swagger + psych up.
:Steelix:3 is not a lot better without access to EQ.
:Exploud:4 is a pokemon i suspect is often overrated for its coverage. Fundamentally its still slow and underwhelming offensively. If :Exploud: is good at something its being in 3hko (and not 2hko) range for the typical power level seen on set 2 and 3 pokemon.
:Donphan:4 is probably underrated although while its trade abilities are good, it is still slow and its still weak to some of the most common types in Factory.
:Skuntank:4 is actually fine. Fast explosion from a real attack stat is instant utility, but poison/dark physical coverage is also welcome. The EQ weakness is unfortunate but :Skuntank: can survive most non-STAB EQs.
:Gengar:4 is probably a top 10 factory set overall and just exceptional. The AI has no understanding of avoiding the player's Destiny bond; at worst :Gengar:4 can trade itself for a nasty threat.

Destiny bond leads are almost always exceptional and get better with teammates that work as a defensive check to common things that force them out.
:Gengar: + :Skuntank: is a lot better than it might sound at first.
The choice for the 3rd is harder. It's never :umbreon:. I am ultimately distrustful of :Exploud: and :Donphan: working well on a team that probably wants to hard-switch 2 or more times per battle. :Steelix: is a worse set overall but functions better on this team.

Team: :Gengar:4/ :Steelix:3/ :Skuntank:4

Battle 1: (Tier 5) :Drifblim:/ :Ursaring:/ :Houndoom:
:Drifblim:2 is a contender for the worst factory set.
Dbond on :Ursaring: was not completely necessary as :Skuntank: actually outspeeds Quick Feet :Ursaring:. This is something I should have properly considered. Regardless, with Lum berry Dbond is still completely safe.
1hp :houndoom: could ohko :steelix: with a crit reversal, (and can set this up via endure anyway) - so the main thing to avoid here is to not explode into sash to deny the AI this chance.

physical :Houndoom: is not a good pokemon but Focus Sash largely makes up for it, and I'm keen to be rid of :Steelix:3 before it embarrasses itself. Focus sash is a better item on a faster team, where often the only thing required of :houndoom: may be to KO a single pokemon below 40%.
(Focus Sash also prevents other focus sash pokemon from appearing which is nice)

Team: :Gengar:4/ :Houndoom:2/ :Skuntank:4

Battle 2: (Tier 5) :Lapras:/ :Sceptile:/ :Ambipom:
:Lapras:2 is standard Frontier nonsense. It's even more annoying in the factory as you hate to face it but also do not want to swap for it.
In gen4, OHKO moves go to a +1 internal AI score 25% of the time, giving them an overall chance of being picked around 44%. Lum berry does block other moves though, and the best overall chance is just to go for the tbolt 2hko.
:ambipom:2 is an interesting gimmick set that the AI fails to use properly. I flip flop on how to rate this, although the inconsistency usually means "bad".
:Sceptile:2 is a joke although double team is never that funny

I decided to go for :Ambipom: over :Skuntank: because of the even better speed tier. I don't know if I fully endorse this decision in hindsight as :Ambipom: usually has to commit to staying in to be useful and the team doesnt really care about twave support.
It is true that a lot of strong physical sets that could force :gengar: out also would beat :skuntank: but really this is why :houndoom: is on the team anyway.

Team: :Gengar:4/ :Houndoom:2/ :Ambipom:2

Battle 3: (Tier 6) :Typhlosion: / :Dugtrio: / :Porygon2:
:Typhlosion:3 is not something I would happily trade :gengar: for but also :Houndoom:2 isnt guaranteed to beat it even with tfang + 1hp reversal.
I switch to :Ambipom: here without fully considering the possibility of it fainting to blitz as adamant :Typhlosion: isn't a common consideration. Blitz does reveal itself to have an OHKO chance
The good news is that despite this, :gengar: just needs :typhlosion: to take recoil in order for sludge bomb to finish it off
:Dugtrio: has something like a 6% chance to survive shadow ball but I do not trust :houndoom: to win a blind 1v1 and 94% to kill and then likely get a free dbond win seems worthwhile.

:Ambipom:2 was uniquely bad in this trainer but I'm not sure that makes it a mistake to take.
Regardless :Porygon2:-3 is a massive upgrade over it.

Team: :Gengar:4/ :Houndoom:2/ :Porygon2:-3

Battle 4: (Tier 6) :Blaziken: / :Slaking: / :Venusaur:
It really feels like whenever I say fire types are bad, one shows up with a decent matchup.
Tbolt on :blaziken: is a small optimisation to deny blaze and maybe get PRZ. :Blaziken: is more likely to shadow claw here, but the damage range overlaps with blaze kick.
Switching to :houndoom: was the best option here as 1hp reversal obviously sets up a :gengar: KO. It's easier to spot this with some hindsight and I'd put this into the "trying to make the more advanced strategy work" bucket when it comes to intentionally trying to dodge blaze.
:Slaking:3 is actually faster than :Gengar:4 and could have picked any move except return. Saving :Gengar: to dbond seems very worthwhile. In hindsight I prefer going :Houndoom: here instead of sacking :porygon2:.
The last pokemon being :Venusaur: meant there was never a real threat of losing as all 3 pokemon should either win (or "tie") that 1v1.
This match was sloppy and indicative of the kind of errors made when not fully paying attention and playing a bit too fast.

:Slaking: is actually an exceptional Factory pokemon. The question is whether :Houndoom: has overlived its spot and the answer is yes. :Slaking: is faster than :houndoom: and doesnt need focus sash to avoid dying to most crits. STAB LO Return hits much harder than 1hp reversal anyway
Ideally :Slaking: would be the lead but it's a small complaint for an otherwise strong team.
The dual-normal weakness is fake here, a fighting type punish is both rare and probably a fallacy as jolly :slaking: hits so ridiculously hard. Be aware of when its fine to double on types.

Team: :Gengar:4/ :Slaking:3/ :Porygon2:-3

Battle 5: (Tier 5) :Arcanine:/ :Aerodactyl: / :Armaldo:
Despite the switching required this was a pretty straightforward win.
The only weakness of the team really is just :gengar: being stuck in the lead position.

Preview shows fighting speciality which is just enough for me to take :Arcanine:. While :Slaking:/:Porygon2:-3 are fine into most fighting types there is no reason to risk a matchup like QC :Machamp:3. :Arcanine:2 is a pretty big upgrade over the original :houndoom:2 anyway.

Team: :Gengar:4/ :Arcanine:2/ :Porygon2:-3

Battle 6: (Tier 5) :Breloom:/ :Ninetales:/ :Poliwrath:
very straightforward win.

I decide to take :Ninetales: over :Arcanine: for the final battle. The reasoning here is pretty much exactly that Overheat is a safer 1v1 option due to no recoil, and :Ninetales:2 is also slightly faster than :Arcanine:.
I again want to call out "Mixed attacking" as overrated. If :snorlax:4 appears then :gengar: can dbond it. :Blissey:4 is not a threat to any team as it can be trivially pp stalled. :Regice: isnt possible. :Milotic: and :Tentacruel: still lose (or "tie") to both :gengar: and :porygon2:
:Poliwrath:2 is somewhere on the radar as a potential switch but it's far too slow.

Team: :Gengar:4/ :Ninetales:2/ :Porygon2:-3

Battle 7: (Tier 7) :Vespiquen: / :Quagsire: / :Ursaring:
Despite the deliberation this was a simple :gengar:4 win.
(2 hours 18 mins if the timestamp link fails)

(Smart AI, Tier 6-7 opponents + Thorton 2 on battle 7 (31IV set 4 Pokemon + Legendaries ), draft is 1x tier 7 and 5x "Tier 8" (31iv any set))

Round 7 is slightly easier than Round 8+ although you really are hoping that your randomised "Tier 8"s here don't suck so you aren't at too big of an IV disadvantage.
You "should" (on average) find a pretty good team by the time Thorton2 comes, as on average you'll have around 3 battles against tier 7 opponents with set 4s to choose from, plus occassionally there are some good set 3s on the tier 6s.

Draft: (Tier 7 / 24iv) :Marowak:4
(Tier 8/ 31iv) :Raikou:4/ :Starmie:1/ :Weezing:3/ :Lucario:2/ :Crobat:1

:Raikou:4 is absolutely a Factory superstar and also an exceptional lead to punish anything underpowered.
:Marowak:4 is high variance but ultimately worth avoiding. it's simply too slow and too unreliable. Best case scenario is usually that it KOs a single pokemon and faints to the next.
:Crobat:1 is ok at best.
:Starmie:1 is a strong support option with great speed, a good item and acceptable moves.
:Weezing:3 is actually something I rate surpisingly highly. Psychic coverage is unusual in comparison to regular PvP singles. It has enough attack to make explosion work and at worst usually trades itself. It's better on stronger teams though, and isnt a carry.
:Lucario:2 is unfortunately the worst :lucario: set as 20% miss moves are infamous for a reason. However it's still pretty quick and still has amazing typing. Its a set you dont want to take but will often be correct to take. The trick to using it is to manipulate the game so that you don't always put the outcome on a 20% chance. It's better in the back for this reason but can work fine as a lead.

:Raikou:4 is an autopick, and its best friend in this draft is :Starmie:1.
The backup pick is definitely :weezing: vs :lucario:. :Weezing: has a higher floor but lower (much lower..) ceiling.
Ultimately I decide that because :Raikou:/:Starmie: is well above average, against the "nightmare fuel" sets (Band :Dragonite:4, etc) i would rather have :lucario: and I would also rather have :lucario: in a pinch situation due to its superior speed.
At this stage the hope is that :Lucario: can be swapped for something better.

Team: :Raikou:4/ :Starmie:1/ :Lucario:2

Battle 1: (Tier 6) :Meganium: / :Lanturn: / :Porygon-Z:
After the FRZ i briefly consider switching and CM. This would be a very bad idea though as overgrow leaf storm is a considerable punish and it would be dumb to throw away so much of :Raikou:'s HP needlessly.
:Lucario: missing Cross Chop on :lanturn: was a real dread moment. Luckily :Raikou:4 is strong enough to not have to gamble on illuminate vs volt absorb.
:Porygon-Z:-3 is the bad no-STAB set although it did reveal download at least.

I decide 1 critical cross chop miss is enough for the round though and go with :Porygon-Z:. Dropping :Starmie: is almost an option here as its set 1 isn't incredible.

Team: :Raikou:4/ :Starmie:1/ :Porygon-z:-3

Battle 2: (Tier 6) :Heracross:/ :Abomasnow: / :Crobat:
CM turn 1 is necessary to avoid a Swarm Megahorn OHKO. However Stone edge gets its 25% +1 bonus for being high crit, and then wins the tie with megahorn to be selected ahead of it anyway. Still want to underline how much of a mistake turn 1 extrasensory would be though.
Possibly even better is turn 1 shadow ball which was also a consideration. But the "reward" of having +1 :Raikou: is a potential free win.
:Abomasnow:3 dodges the 2hko roll fairly often at only +1 so CM was a reasonable although slightly greedy click which I am punished for via a FRZ.
I spend a while here deciding if this is Psychic range or if I have to waste Download and signal beam. Later in the streak I'd probably get out a pixel counter but here I go with my gut which says this is Psychic range.

:Heracross:3 is "pretty good" but is inconsistent for the same reason as :Lucario:2.

Team: :Raikou:4/ :Starmie:1/ :Porygon-z:-3

Battle 3: (Tier 6) :Slowbro:/ :Rapidash:/ :Dusknoir:
Horn drill brightpowder :rapidash: is another classic frontier set, but I avoid any huge bad luck disaster.

I strongly consider :Slowbro:3 (specs) here but no water STAB and a terrible speed stat are enough of a deterrent. Plus I'd rather put the match on :Starmie: anyway.

Team: :Raikou:4/ :Starmie:1/ :Porygon-z:-3

Battle 4: (Tier 7) :Empoleon:/ :Dugtrio:/ :Tangrowth:
Honestly, I can't remember if I genuinely wanted to test arena trap or if i had just forgotten about shuca berry here. Either way the chance of losing was very low after making it a 3v2.

Rock in preview makes :Empoleon: a candidate swap but I do not really want to swap :Porygon-Z: for it.

Team: :Raikou:4/ :Starmie:1/ :Porygon-z:-3

Battle 5: (Tier 7) :Aerodactyl:/ :Probopass: / :Flareon:
Aero4 (band) is scary.
Decomp knowledge helps a lot here. Stone edge can OHKO as well as EQ (the AI will misread shuca berry), but Stone edge can be elevated above it as a kill move as it is high crit.
This means that :Starmie: is not the safe switch it may first appear to be. (:Starmie: is faster than :Aerodactyl:)
:Raikou: could definitely stay in and simply thunderbolt, risking a crit or stone edge roll.
What is safest is to just sack :Porygon-Z: (if it locks into Stone edge :Porygon-Z: could actually win with a single miss) and then go :Starmie:.
Probo4 looks like a joke but with LO it actually manages to get the OHKO roll on :Starmie:
:Flareon:4 last is very concerning as its one of the few factory sets almost favored into :Raikou:4 (the AI increases the priority on giga impact at lower HP).
I can't even tbolt-cm-tbolt to disincentivise giga impact as the first tbolt does too much. I consider CM-Extrasensory-tbolt but I suspect this is a roll and intuitively it feels obvious that the roll will unfavorably overlap with giga impact's score threshold of 60%HP.
Overall the best option seems to be to go for tbolt into either extrasensory+tbolt for the cumulative crit/flinch/Prz chance on a low roll, and hope that :Raikou: can edge out the chance of dodging an attack - or hope that the first tbolt is a good damage roll to get the 2hko.
I do get "lucky" and crit the first tbolt, although :Raikou: is probably overall favored to win this 1v1 perhaps 75% of the time if I correctly react to the first tbolt roll.

It's hard to say if this battle was played poorly without hindsight.
Switching to :Raikou: against :Probopass: to go for CM into 2x tbolt is not safe, and neither is 2x tbolt as after 26hp damage from life orb, :Raikou: can still miss this range.
I haven't done a full breakdown but it seems intuitively obvious that :Raikou: beats more things in a 1v1 than :Starmie: does.
Perhaps :Raikou: is supposed to stay in on turn1. If it was castle or Arcade and I knew the last 2 mons i would have done this, as Pory can easily get enough chip in alongside LO for the surf OHKO on probo.

aero4 is very strong regardless and what the team is currently "missing" to complete a fast and hard hitting trio with decent defensive switch checks. It's also an EQ pivot.

Team: :Raikou:4/ :Starmie:1/ :Aerodactyl:4

Battle 6: (Tier 6) :Blissey:/ :Ninetales: / :Sceptile:
I'm glad I do not have to enter a CM war with :raikou: against :blissey:.

:Blissey:3 is often overlooked in how strong it is in earlier rounds where you know if the set you are facing is physical or special.
That said, Thorton can have variable sets if he picks from the "legendary" tier which makes :Blissey: unreliable, and :blissey:3 does lack recovery so its more of a "I'll come in and beat this one thing for you" kind of deal, as its rare to want to save :blissey: on the switch in after its KO.
:Sceptile:3 is not dreadful but :Aerodactyl:4 is in a different league

Team: :Raikou:4/ :Starmie:1/ :Aerodactyl:4

Battle 7 (Thorton): :Alakazam:4 / :Gastrodon:4/ :Rapidash:4
:Raikou:'s best chance vs zam is to CM once. It can still be 2hko'd I am sure, but in that situation :aerodactyl: can just come in and OHKO anyway - so there is no utility in getting safe chip damage in with shadow ball
:Gastrodon:4 and :Rapidash:4 are weak sets in comparison to what Thorton can field from and this match was hard to lose, even if :Alakazam: beat :raikou:.
From Round 8 onwards, "elevations" no longer exist. This is factory in its purest form: any Tier4+ pokemon can appear on opponents teams and in the player's draft. They are all 31IV

With the exception being Tier 7 trainers (Trainers 201-220) - these will always have 24IV set 4 pokemon. 7IVs isnt that much at level 50, and the power of set 4s often compensates for it - although it is nice to not have to scout out what set it is.

(Yes, roughly 20% of post-49 trainers are 24iv + set4 and roughly 80% are 31iv + random set)
(2 hours 45 mins if the timestamp link fails)

This is where the run starts getting a bit more serious and where I want to focus more as while Factory is "easier" than it used to be, I would still say getting to 49 is optimistically like a 50-70% chance and a few hours of play.
From here on, all pokemon are 31IV unless its specifically a 24IV opponent (set 4 guaranteed). This makes damage estimates from base stats a lot easier.
There is also absolutely no reward for swapping from now on.

Draft: :Gallade:4/ :Breloom:2/ :Froslass:3/ :Golem:2/ :Exploud:3/ :Crobat:3

:Breloom:2 and :Golem:2 are simply bad.
I dislike :Exploud:3 as a slow life orb pokemon without a strong primary STAB button.
:Froslass:3 is one of the best leads in the entire factory and an autoinclude.
:Gallade:4 has its problems but is fundamentally a solid pick
:Crobat:3 is not great but at least its fast

It's worth briefly checking to see if the obvious team makes sense. :Froslass:/:Gallade:/:Exploud: would have been the alternative and for many reasons is worse.

Team: :Froslass:3/ :Gallade:4/ :Crobat:3

Battle 1: (24iv/ set4) :Quagsire:/ :Garchomp:/ :Victreebel:
:Garchomp:4 is an incredible pickup although -7 IV in speed is a bit sad as it lands on the wrong side of the 152 benchmark.
This :Garchomp: exposes the problems with pokemon like :Gallade:4 who are 255 (atk stat) / 255 (spDef stat) - which is that if they are outsped by something hitting on the other side of the spectrum, they just drop.
The factory almost always pumps more EVs into the better defense stat which makes this problem "worse" (most infamous example of this is surely :blissey: although :gallade:'s 50 base difference between its defenses is another good one)

The choice on what to drop between :Gallade:4 and :Crobat:3 is not that close as :Gallade:'s typing and coverage synergise quite a lot better.

Team: :Froslass:3/ :Gallade:4/ :Garchomp:4

Battle 2: (31iv/ any) :Electivire:2 / :Sceptile:2 / :Tentacruel:2
This battle demonstrates the added difficulty of 49+ well. 2/4 :electivire: click an electric move on turn 1, then the other 2 click iron tail and fire punch.
On average, the :Garchomp: switch is rewarded and I get to (almost) freely scout. But its the iron tail set and it crits.
:Sceptile: coming out next strongly suggests Set 4 by AI-switch decomp logic. The safest thing to do in that case is sack :garchomp: to safely get :gallade: in for CC.
But set2 is revealed instead, and so the better play becomes to go :froslass: to reduce the need to switch for a potential final pokemon dbond.
This also reveals that the last pokemon has a move "as bad or worse" as giga drain against :Garchomp:. i.e. no hyper voice/ ice beam/ close combat/ explosion/ etc.
Such a case is very rare, and when :Tentacruel: comes out I immediately know it must be set2 and that it tied with :Sceptile: for strongest damaging move (giga drain)

Of course, the battle is easily won without this level of critical thinking but its easy to imagine how in closer games this level of scouting helps.

None of these pokemon are swap options.

Team: :Froslass:3/ :Gallade:4/ :Garchomp:4

Battle 3: (31iv/ any) :ursaring:1 / :Cresselia:3/ :Blastoise: ?
Normally :Froslass: would not trade itself for :ursaring:, but I have more faith in :Garchomp:4 to potentially carry and there isn't really a safe switch into an unknown :ursaring: anyway.
After dbond the AI continues its logic as if :froslass: was still in. This means that when :cresselia: comes out its slightly more likely to be the shadow ball set (3) since this is the best anti-:froslass: set.
:Garchomp: survives set 4's ice beam - and even if it didnt a switch here would be a major risk. The AI crits :garchomp: (a disaster to be sure), leaving an uncomfortable ~55% :Gallade: vs an unknown 3rd.
:Blastoise: comes out.
It's very unlikely to be set 3 as :Ursaring:'s aqua tail would do a lot more than :Ursaring:'s Shadow ball against :froslass: (this is how switch logic works internally).
Set 2 is also less likely as even waterfall usually will outdamage shadow ball from :ursaring: against :froslass:.
Note that :Gallade: simply loses to set 2/3 anyway without luck.
Set 4 will actually not beat :Gallade: in this scenario as - even if they kill - Hydro cannon and Hyper beam score lower than ice beam on the internal AI logic. Meanwhile leaf blade will 2hko.
Set 1 is basically the only interesting case as it's chance of selecting mirror coat is quite high, and hydro pump can OHKO :gallade: but only in torrent range.
Unfortunately the Psycho cut + leaf blade KO is not guaranteed, and leaf blade itself is not guaranteed to give torrent anyway.
So after mapping out all of those possibilities, the optimal move is just super effective leaf blade anyway.

:Cresselia:3 is incredible at switching into something, taking the hit and then outspeeding and getting a 2hko while avoiding the 3hko. It does other things well too but that's the main party trick.

Team: :Froslass:3/ :Cresselia:3/ :Garchomp:4

Battle 4: (31iv/ any) :Houndoom:3 / :Gengar:1 / :Lopunny:2
The :houndoom: sets that do more damage to :garchomp: are also the sets that are outsped and OHKO'd. The annoying reveal would be :houndoom:1 plus counter (I do not have time to safely break sash via crunch), although at least this would mean that the last 2 pokemon cannot be dark.
Payapa :gengar: is a very deadly follow up to something that forced :garchomp: in. the outrage crit didnt matter but a shadow ball crit from :gengar: would have been gg and a very rare case of a double-ghost weak team actually getting punished for the ghost weakness.
Once again I sympathise with the (now known to be wrong) old view that the factory "cheats". I just give up :gallade: for a clearly superior take, and instantly payapa :gengar: appears (something night slash :gallade: can probably handle). At times it almost feels intentional.

Team: :Froslass:3/ :Cresselia:3/ :Garchomp:4

Battle 5: (24iv/ set 4) :Drapion:4 / :Altaria:4/ :Gallade:4
This was another rushed decision on turn 1. :Garchomp: "feels" safe but a sniper night slash crit almost drops it.
Another reminder that the factory is constantly ready to punish the slightly mistake or inaccuracy.
Battle was otherwise very straightforward

Despite how strong :froslass:3 as a lead is, both :Drapion:4 and :Gallade:4 are very strong sets. :Gallade:4 is somewhat discounted for the type overlap, but :Drapion:4 is a real possibility. I think without :Garchomp:4 in the back, I probably drop :froslass: for it.
With :Garchomp: in the back however, I would much rather just reduce the variance in what can go wrong in terms of an enemy sweeper set lead somehow doing too much either early or late.

Team: :Froslass:3/ :Cresselia:3/ :Garchomp:4

Battle 6: :Marowak: ? / :Honchkrow: ? / :tentacruel: 1
This is about as easy as it gets in terms of bad enemy teams. :Garchomp:4 alone likely can 1v3 here. As-is, :froslass: easily cleans :marowak:, :honchkrow: gets prioritised due to its dark stab move and then easily is beaten too.

:Tentacruel:1 could work ok as a lead since the team has basically perfect coverage on its weaknesses while it is actually faster than :froslass:3. it's bulkier and denies most problematic pokemon via icy wind.
That in itself is the problem though, destiny bond doesnt have a 5% miss chance and I can think of enough kind-of bad matchups (mainly opposing waters but also something that hits super hard like :tauros:4). In the end it feels safer to stick.
I am also slightly off-put by the Psychic hint for the upcoming trainer.

Team: :Froslass:3/ :Cresselia:3/ :Garchomp:4

Battle 7: :Bronzong:3/ :Typhlosion:4 / :Gardevoir: ?
:Bronzong: is a rough pokemon to see on turn 1 and a potential TR is very bad. This not only potentially denies a dbond but also flips a reasonable number of :Garchomp: matchups.
Hypnosis is completely fine as it can be switch stalled out of PP completely by keeping :froslass: as the sleeping pokemon (preventing signal beam).
:Bronzong:4 threatens an OHKO with Iron head.
Overall :Bronzong: deserves enough respect to just immediately remove via dbond. :Bronzong:3 doesnt offer any problematic moves.
:Garchomp: works well with lead :froslass: as an ice type will almost never be the AI's choice to bring in after it faints which further limits the % of things with a chance to beat it.
:Typhlosion: and :Gardevoir: are both not on this list and are removed.

:Cresselia: didn't actually do anything but it would have been useful in all of these matches if :Garchomp: wasnt around.
:Garchomp:4 really is the factory God in the hands of the player and the AI. I don't think its an exaggeration to say it may on average be favored in random 1v3s, and realistically in a 3v3 it just needs a single teammate to reliably cover whatever its one bad matchup is.
(20s in)

First hint: Ice

Draft: :Dewgong:4, :Machamp:2 (NO Guard), :Skuntank:1 (Aftermath), :Porygon-Z:1 (Adaptability), :Drifblim:3, :Hippowdon:1

This draft is "problematic" but workable. No matter what I pick adjustments will definitely have to be made to avoid a loss in the next 7 battles. Most concerning is a lack of anything both fast and (power-wise) acceptable.

:Porygon-Z:1 is basically an instant pick but it is not good in the lead slot
:Skuntank:1 and :Drifblim:3 are both discounted as options pretty fast.
:Skuntank:1 should be a good set but sucker punch is a horribly unreliable move to pick unless you're literally about to faint. I would much rather gamble with naked Thunders or Stone Edges.
:Drifblim:3 toes the terrible line against an ice specialist and would be a nonsense pick.

:Hippowdon:1 is the best pokemon after :Porygon-Z: and an argument could be made for taking it even into an ice specialist despite the fact that pretty much every ice type in the game will beat it except the terrible ice types which i dont care about.
:Machamp:2 and :Dewgong:4 operate on a very similar power/ utility scale post-49. Pre-49, :Machamp: edges it out because there is no need to scout for physical sets for counter.

:Dewgong:4 stands out as the best pick into ice though, with resto chesto and surf probably being enough to safely 1v1 without worrying about freezes. Since I wouldn't want to keep it for long it feels like a natural lead pick.

:Machamp:2 vs :Hippowdon:1 is something I was really not sure on. My typical advice is to not over-specialise for a battle 1 opponent and take hippo, but I am sufficiently scared of a pokemon like :Regice:4 or even :Lapras: showing up and decide the :Dewgong:/:Porygon-Z:/:machamp: draft is a fixable core.

Team: :Dewgong:4/ :Machamp:2/ :Porygon-Z:1

Battle 1: :Meganium:3 / :Glalie:2/ :Walrein:2
(Unknown) :Meganium: is a small disaster as a lead. :Dewgong: not only fails to damage it without potentially giving overgrow, but I do not want to give any time to the screens, weather or stall set. On top of all of this its also faster.
:Meganium:3 is a lot less scary than it looks because of how spatt lowering moves are delisted as "best damage" in the decomp.
After wring out crits :Dewgong:, :Meganium: is (critically) not in overgrow, but will definitely see the KO anyway. It is important to preserve :dewgong:4 as it has exceptional matchups into most ice types (this was the entire reason it was drafted).
Anyway, guaranteed non-overgrow leaf storm means porgyon-Z can come in on anything except a crit.

Unknown :Glalie: in second is a small surprise as most of its sets have low-base power moves. The highest would be explosion (set 2) and by some weighting this is also the most likely :glalie:, although Blizzard (set 4) is also reasonable.
Regardless, :Porygon-Z: outspeeds all :glalie: sets and should be safe at 123hp. :Glalie:2 is revealed and goes down.
An alternative strategy (:Dewgong:) would be viable if it wasnt for :Glalie:4. Sheer cold is enough of a punish and it's slightly more likely to be picked than blizzard.

:Walrein: is in last and (somewhat rarely for the 3rd reveal) we know nothing about it. We know its strongest move is weaker than Explosion (well duh). We cannot discount any item.
Set 3 (OHKO spam) is not actually that bad despite a slower no guard :machamp: because of sash. In different circumstances or with different earlier decisions it could be a problem though and this is mostly a highlight of the importance to position the team optimally even when given good matchups.
Knowing that doesnt change anything though, and tri attack reveals set 2. Set 2 has no chance to win.


:Machamp:2 >< :Meganium:3 is a small consideration as a swap. I do not view this as a significant upgrade though. While screens are powerful, they are poor insurance against bad luck (crits) compared to a focus sash.
I strongly dislike having to lead :Meganium:3 as its a lead which is "forced out" by way too many opponents, and this team isnt good enough to handle that kind of turn 1 momentum loss.

Team: :Dewgong:4/ :Machamp:2/ :Porygon-Z:1

Battle 2: :Absol:3/ :Raichu:4/ :Starmie:4
Lead :absol: is actually quite scary. The obvious / intuitive play of bringing :Machamp: has a lot of problems with not only swagger/ crit but also just a straight up damage race loss against :Absol:4.
However I know from the decomp that superpower and giga impact are both heavily discouraged with :absol: at 100%.
This transforms the best play into staying in. Signal beam does not meaningfully outdamage ice beam, and frz would be a better random event than confuse.
:Absol: 3 is revealed and after a "free" (chesto) heal, :dewgong: wins while staying relatively healthy.
As mentioned in the video, I was surprised at the night slash damage from 3 (was not expecting a 2hko roll from the jolly set). Small reminder to always check rolls that may be close.

Against unknown :raichu: the best thing to do is definitely just sack :Dewgong:. :Dewgong: is not fast enough to reliably be useful later on.
There's no way its worth trying to save it just to see :raichu: either reveal focus blast or to get paralysed.
Note that a better (or at least a faster) lead would probably make me go :Machamp: early.
After :Raichu:4 is revealed, :Machamp: safely comes in to remove it via EQ.

:Starmie: last is a disaster. But the AI reveals one of its worst flaws which is to go for any kill that it "simualted". When checking to see what moves kill, it rolled a max damage surf and so went for surf.
But then the actual roll was lower, giving :machamp: a turn.
This is "lucky" but without crits :porygon-Z: wins the 1v1 against :Starmie:4 anyway.


:Starmie:4 is a top 10 factory set and a huge improvement to the team.
The question of what to replace (:machamp: or :dewgong:) is harder. The natural answer (:dewgong:) feels right and after running through a quick list of sets that would actually force a :starmie: switch I go with this.
It's worth underlining that this is a tough decision because anything that can force out :Starmie: on turn 1 is a disaster as :Machamp:2 is only really useful while it has a focus sash, and :Porygon-Z: struggles in a lot of speed check matchups.
For this reason the team is still a swap away from true safety and there is an argument to be made that :Hippowdon:1 should have been taken, even without this hindsight benefit.

Team: :Starmie:4/ :Machamp:2/ :Porygon-Z:1

Battle 3: :Manectric:3 / :Walrein:2 / :Vespiquen:4
:Manectric: is one of the few things that does force out :Starmie:4 which is annoying. Luckily :Machamp: does enter a tank role for once, exposing weaker attackers.
EQ finishes :Manectric:.
:Walrein:2 comes in which is quite heavily suggestive of set 2 (since :manectric: is physical and set 2 has the highest base power move).
The AI does not double Avalanche's power internally when scoring moves so it does not see the kill here.
Curse is a fine outcome since :Machamp: is now faster and can survive EQ, which could lead to :Machamp: "winning". :Walrein: does not EQ and :Machamp: falls.
The faster and more frail :Starmie: is a better switch in for the final pokemon.
:Vespiquen: is a terrible pokemon.

I would never normally consider one of these but the Flying hint is on some level concerning. :Aerodactyl:4 and :Zapdos:1 among others would be extremely threatening leads.
After running through the actual threatlist quickly, it becomes obvious that none of the swap options even improve it that much.

Team: :Starmie:4/ :Machamp:2/ :Porygon-Z:1

Battle 4: :Yanmega: ? / :Aerodactyl:3 / :Skarmory:3
Lead :Yanmega: only has a chance if protect + speed boost happens (set 1 + 50% right ability). It doesnt happen.
Non-lead :Aerodactyl:4 is a lot less threatening as there isn't as much pressure on me to preserve :starmie:. If it is :Aerodactyl:4 then :machamp: can come in.
Aero3 is revealed instead and falls to surf.
:Skarmory: as the last pokemon offers literally 0 threat and as such its optimal to scout it for eliminating item options later. Using the weaker surf reveals set 3.

My opinion at the time of the write up disagrees with the next swap. A "known" :Aerodactyl:3 is a lot better than an unknown :yanmega:, although it is true that they're both better than :Machamp:.
The comment i make on stream (pls no :blissey:) isnt even the reason for this, :Aerodactyl:3 is just a more reliable backup.

:Starmie:4/ :Yanmega:3/ :Porygon-Z:1

Team: Battle 5: :Kangaskhan:1/ :Roserade:4/ :salamence: ?
I check :Yanmega: and set 3 is revealed, which is something of a punishment and further reason to go with :Aerodactyl:3.
:Kangaskhan: is actually a problem. While set4 will not giga impact, the rest all threaten to do "too much" damage to an incoming :yanmega:.
:Porygon-z: as an option would be better if it was actually faster than any :Kangaskhan: set.
The decision to surf vs psychic is close. Against set 1 it would definitely be better to psychic, but against every other set the small increase in power from surf can matter.
Set 1 (salac) is a punish because it can knock itself into salac range via double-edge (which would be more likely than crunch). However, salac range _always_ overlaps with a HP range where :Kangaskhan: would kill itself attacking :Yanmega:.
As-is, salac does not activate and the match continues.
Some :roserade: sets (1/2) can survive STAB psychic while the others (3/4) definitely faint. I reason that :Starmie: is valuable enough to keep alive to the degree where i'm willing to lose :yanmega: for "nothing" if it is set 3 or 4.
:Roserade:4 being revealed means that :Yanmega: did "die for nothing" but I think this remains the best play.
(Note that in the video I did know it was :roserade:4 after seeing 2x shadow ball but since the items overlapped i likely ignored).
:salamence: is out last and as mentioned in the video, I don't have the luxury of messing around to check out the set (set 2 considers the speed tie to be "slower" and will DD).

:Yanmega:3 is leaving no matter what since its worse than all 3 swap options offered here.

I go with :salamence: in the end but :Kangaskhan:1 was a serious option. The worst :salamence: set (1) still has intimidate and avoids stacking a fighting weakness.

Team: :Starmie:4/ :salamence:1/ :Porygon-Z:1

Battle 6: :Lopunny:2 / :Leafeon:4/ :absol:1
Unfortunately, it is :salamence:1.
:Lopunny:2 is annoying but that's about it so long as you respect Focus punch.
On turn 3 after switching for natural cure we see the very rare resist-switch on surf. :Leafeon: comes in but is intimidated since :Lopunny: was faster.
:Leafeon:4 is revealed via the :Salamence:1 damage roll against it.
:Absol: and :lopunny: offer little resistance.

Next hint is water.
:Leafeon:4 is well above average for a grass type but its still a grass type and has a hard time convincing anyone is worth a swap.
The water spec doesn't really change my mind on this.

Team: :Starmie:4/ :salamence:1/ :Porygon-Z:1

Battle 7: (24iv) :Blastoise:4/ :aggron:4/ :Milotic:4
The only interesting thing about this fight is what to send in after sash removes :starmie:.
Keeping intimidate as a switch option is usually good but scary face support means :porygon-Z: really shouldn't lose to much either. In the end i preserve intimidate.
A 24IV :milotic: cannot realistically stall out tri attack, although it attempts to do so. but (even with lum) the odds are too heavily stacked in my favor to find either double status or a single crit.

As a quick postmortem, :Leafeon: would have been good in battle 7 to avoid any chance of losing. But I could equally have lost to :gyarados: or :Ludicolo: and been annoyed i took it at all.
 
(cont..)
(30s in)

Draft: :Staraptor:2, :Glaceon:3, :Jolteon:2, :Blissey:2, :Poliwrath:1 (damp), :Cresselia:4
First hint: Rock

:Cresselia:4 is a very strong set and an instant pick.

The rest are all bottom-half in terms of post 49 sets and represent quite a serious issue for the draft.
:Staraptor:2 is perhaps the worst of them without any real offensive pressure and limited utility outside of intimidate. Saving or trying to setup endeavor goes wrong too often and too easily, and the team is hangstrung in attempting to play in a way to avoid :staraptor: being involved in a late 1v1.
:Blissey:2 is a gamble. The tl;dr on :blissey: is that since it has 0EVs in defense, it is shaky into physical sets. But is extremely safe into most special sets. The thing is that OHKOing 252hp/ 0def bliss is actually a reasonably high bar. We'd be talking about a 130 base + 130 effective power invested physical set to get a roll.
Counter is a brave click into an unknown though as OHKOs not being guaranteed simply means the AI can choose a status move or could run out some utility move instead like fake out. There will also be pokemon which have 1 or 2 sets that may OHKO :blissey:. On top of all of this is the truth that a critical hit of almost any STAB attack is the end of :blissey:.
pre-49 :blissey:2 is actually a good pokemon because you can reliably know if counter is viable and you will know the enemy's category without scouting.
:Jolteon:2 is more of a meme than it may appear. A 0EV speed :jolteon: loses a lot of its appeal. Having said that, it does have thunder wave and discharge is ok and it probably contributes something.
:Poliwrath:1 is slightly below average and is mostly a slightly slow surf button. This prevents it from being that bad though, particularly into rock.
:Glaceon:3 is too slow to be called average but its not terrible. It is not necessarily terrible into rock depsite the "super effective" matchup rock has.

The decision I make is a compromise; :Poliwrath: can beat a lot of Rock types and is likely the 2nd best pokemon after :Cresselia:4 anyway.
I lead with bissey. The way :blissey: works is that although its semi-reliable as a counter click, it cannot switch in and then click counter. The alternative was probably to lead with :staraptor: and endure-endeavor against most opponents. As mentioned above, this can backfire easily though. I'd rather just click counter on :blissey: into most enemies and :blissey: has the additional benefit of destroying not just special sets but also some of the mega annoying stall nonsense that may be too much for Cress.
:Poliwrath:1 can work for a few games but I am eager to be rid of it. Against rock its probably forced anyway.

Team: :Blissey:2/ :Cresselia:4/ :Poliwrath:1

Battle 1: :Lanturn:4 / :Golem:2 / :Rampardos:2
:Lanturn: (including physical :lanturn:) is something blisey easily beats no matter what.
:Golem: comes in on +6 evasion :blissey:. :Golem: 2,3 and 4 can all OHKO :blissey:, so the clear decision is toxic.
Switching :Poliwrath: would be a mistake when I know the final pokemon is rock type, I'd much rather lose :blissey: than take a hit on :poliwrath:.
EQ hits :blissey: but does not kill, suggesting set 4. But the next fire punch actually confirms set 2. Once again, "dont use superpower at high hp" is proving itself true.
:Blissey: narrowly beats :Golem: due to the evasion boosts and Ramp comes in.
At 4hp the clear decision is to toxic once again. :Cresselia: finishes it off.

Hint: Ground

Ground is one of the more concerning type hints. I decide that since its only match 2, fixing the team takes priority and a lead :lanturn:4 can at least pivot out into levitate cress.
:Lanturn:4 is far from an ideal lead and I'm probably stuck with it.
STAB EQ from things like Rhydon are on the shortlist of things :blissey: literally cannot deal with. I'd also rather stick with the water type going into a ground specialist.

Team: :Lanturn:4 (illuminate) / :Cresselia:4 / :Poliwrath:1

Battle 2: :Steelix: ? / :Mamoswine:3 / :Typhlosion:2
Between QC and passho, there is risk in :Lanturn: staying in. But its a ground spec and surf is, on average, a high value outcome.
Going into :Cresselia: isn't a good option since cress doesnt beat :steelix: down fast enough, and :steelix:3 would be unpleasant.
:Poliwrath: is an option but runs into similar problems as :lanturn: with QC and Passho.
Deciding that :Poliwrath: and Cress are obviously better keeps into pokemon 2 and 3 (if i know at least one is ground), I just surf and :Steelix: does drop.
:Mamoswine: comes in and literally every set will outspeed and OHKO :lanturn: with EQ. in other words EQ is completely guaranteed and the AI will never choose another move no matter what the set.
I send :Cresselia: in with the intention of maybe going to :poliwrath: later for vacuum wave.
:Mamoswine:1/3 is revealed via Hail which is enough to keep :Cresselia: in for at least another turn.
:Mamoswine:3 is then revealed and :Cresselia: is able to finish it off.
As mentioned in the video, :Typhlosion: has no hope of winning from this position as even set2 doesnt get past vacuum wave, plus set 2 has no hope of KOing cress without locking itself into a recharge turn.

Hint: -

:Mamoswine:3 is a considerably better pokemon than :Poliwrath:1. This is, however, a case where team composition does trump raw "pick the best 1v1ers". The utility of a 2nd water and particularly priority outweights the fact that :mamoswine:3 is slightly better.
This is a "close" decision though.

Team: :Lanturn:4 / :Cresselia:4 / :Poliwrath:1

Battle 3: :Armaldo: ? / :Leafeon:2 / :Honchkrow:1
:Lanturn: easily and safely beats :Armaldo: 2,3 and 4. It's very likely to beat :Armaldo:1. Surf is the clear play.
:Leafeon: comes out next which is limited information beyond the fact that I now know the 3rd pokemon cannot have EQ, explosion or other very strong physical moves vs :lanturn: (stronger than leaf blade).
:Leafeon: is "unlucky" in that sense that its on the shortlist of counters to this team.
I go :Cresselia: since both :Leafeon: 3 and 4 will drop :lanturn: for free and :Cresselia: drops slightly in expected utility if the 3rd pokemon does not run EQ.
:Leafeon:2 is revealed which is probably the worst case scenario as it could maybe win via Grass Whistle. It also "probably" survives ice beam + vacuum wave.
:Leafeon: does instead get a crit though which was close to a disaster. :Cresselia: is just bulky to survive this thanks to 2 leftovers ticks.
Arguably, going :poliwrath: would have been better. This would give :Leafeon: an additional chance to curse again (at which point cress would outspeed), plus if grass whistle doesnt hit then vacuum wave is a free attempt. I feel like despite all of that, staying in is probably still better.

brief postmortem; :Mamoswine:3 would have been better for this fight specifically. Doesn't make it the correct decision overall though.

Hint: -

A known :Armaldo: would be a swap option over :Poliwrath: but an unknown one (particularly if it could be 2) is not worth the risk. :Poliwrath:1 stays and remains the weakest link and most likely cause of a loss.

Team: :Lanturn:4 / :Cresselia:4 / :Poliwrath:1

Battle 4: :Entei:3 / :Marowak: ? / :Bronzong:2
As mentioned in the video, no frontier :Entei: sets run bulk and base 76 modest is pretty much exactly the power needed to OHKO with surf.
:Entei:3 reveals itself but does not use Sunny day. See the decomp for more details but sunny day/ solarbeam are one of the most broken parts of the gen 4 AI.
:Marowak: has no chance against :lanturn: but I would rather not take any chip just to discover its set.
:Bronzong: has no way past Cress unless its set 2. It does end up being set 2.
The 100% safe method of dealing with set 2 is to make sure specifically :Poliwrath: falls asleep (since it will not regularly signal beam :poliwrath:) and waste all of the hypnosis/ dream eater pp.
This kind of exploit on sleep sets is common and recommended despite how slow it is (particularly on console).

Hint: -

:Entei:3 is probably the worst :entei: set but its still better than :poliwrath: and is significantly faster. Its true that matchups :Lanturn: leaves will often be EQs, but Cress is on the team for that.
In hindsight I don't agree with keeping :Poliwrath: again. The decision is still "close" though.

Team: :Lanturn:4 / :Cresselia:4 / :Poliwrath:1

Battle 5: :Slaking:1/ :Lucario:2/ :Politoed:3
:Cresselia: beats all :slaking: sets. While yawn is annoying, its better to actually accept the sleep as its the only way to guarantee that the AI doesnt yawn on the final turn before :slaking: faints.
:Slaking: does get 2 crits off on :Cresselia: which comes close to mattering but not really.
:Lucario: comes out and its more likely to be the physical set 2 or 3 than 1 (4 not possible). :Cresselia: should beat all of these.
:Politoed: last is actually scary. Set 1 and 2 have no real hope of winning.
Set 3 (belly drum) and set 4 (Hypnosis + attacks) both can definitely win though, particularly the belly drum set.
I reason that allowing multiple belly drum turns isnt acceptable (since belly drum -> rest is a huge issue) and actually go straight into :lanturn:.
The reason for this is that if it does belly drum on the :lanturn: switch then EQ is guaranteed before rest, and after it goes rest as i return to cress then I can safely psychic and then proceed to send :lanturn: back into waterfall. This lets :poliwrath: come in first which is vital for :Cresselia:'s chance of winning.
The AI does not belly drum however which means that tbolt safely denies any further belly drum attempts. EQ first would guarantee rest after tbolt, and so :politoed: will never have an opportunity to belly again without sleeping for 2 turns (which cress always can use to deny belly drum).

With hindsight :Entei:3 would have been much more useful due to its speed and the fact that hyper beam does just enough while :entei: (like :lanturn:) also perfectly baits the AI into the 2hko attempt.

Hint: -

It's time for :poliwrath: to go for certain now. :Lucario:2 is a better pokemon than :Slaking:1, but this team does not actually require :Lucario:'s increased damage output.
:Slaking:1 is a much better set than it seems because of how strong yawn is and because of how reliable :slaking: is in general for finishing weakened foes off. Both :Lanturn: and Cress can be guilty of this.
Taking :Slaking: doesnt make winning easier but it does make it harder to lose, which is often what you want in the factory (longer matches increase the amount of opportunities the weaker opponent has to mess up).

Team: :Lanturn:4 / :Cresselia:4 / :Slaking:1

Battle 6: :Leafeon:3/4 / :Nidoking:4 / :Kangaskhan:2
Once again :Cresselia: is effective as a switch in that can safely handle an opponent via alternating moonlights. This is what makes it so strong; its not just how many things it can beat but how many things it beats while maintaining >50% hp.
:Cresselia: also beats :Nidoking:, but unlike the first 1v1 I do not choose to greed any healing at risk of being crit. I trust :slaking:/ :lanturn: to handle most 3rd pokemon and see no reason to potentially lose because of a crit after moonlight (:nidoking:4 is faster than :lanturn:).
:Kangaskhan:2 is not a good set but does land a crit on cress. :Slaking:1 does its thing and yawn -> switch out and :lanturn: can easily handle a sleeping :Kangaskhan:.

Hint: -

I decide to keep faith in the logic behind the previous :Slaking:1 swap, although both :nidoking: and :Kangaskhan: could have been taken.

Team: :Lanturn:4 / :Cresselia:4 / :Slaking:1

Battle 7: :Floatzel:1/3 / :Mamoswine:2 / :salamence:1
:Floatzel: is high on the list of things :lanturn: wants to see on turn 1.
:Mamoswine: appears again and once again its a cress switch to handle it. What's different now is that rain prevents moonlight. I am prepared to take a chunk on :slaking: if needed, but the AI throws away this slight edge by using curse too many times.
I was a bit nervous to psychic on the final turn at the time but in hindsight this easily looks guaranteed.
I did not catch the exact role at the time but in hindsight this is :Salamence:1.
:salamence: is a set that is usually scary to see but :Slaking:1 actually handles it really well (with a teammate assist) with its max hp/def and yawn.

:Cresselia: borderline carried this entire round. I do not think its the best of the best since it's only ever a bit of bad luck away from doing very little, and no slow set can ever truly be the best.
(1:17:39 in)

Draft: :salamence:2, :Shuckle:3, :Roserade:4, :Bronzong:1 (levitate), :Drapion:2, :Honchkrow:4
Hint: -

This is the "WR" draft, meaning that the old singles WR was 74 and this could extend to 77. Because of this going into the draft I have a slight bias towards longer term thinking rather than wanting to over-specialise for a hint. Basically, I'd rather lose on 71 or 72 than 73 or 74 if the reward is a boosted chance of 77.

:salamence:2 is probably *slightly* overrated but its still a strong pick as the player can often either just use it to buffer out 2 physical sets, or sometimes to manipulate a setup. But its worth noting that without outrage or earthquake, it doesnt hit as hard as you would like.
:Shuckle:3 is a contender for the worst set in the factory.
:Roserade:4 is good/average. It elevates itself above most grass types and its good spdef and speed prevent it from losing to most waters even on the switch. Beyond that, its shaky at best.
:Honchkrow:4 is decidedly average but has an unfortunate set of weaknesses; the tldr on that is that it loses to the most dangerous sets and wins against sets that most teams do not struggle with - making it a rare but sometimes important draft.
:Drapion:2 is dreadful although its better than :shuckle:3.
:Bronzong:1 isn't great but it has the best direct synergy with :salamence:. :Bronzong:1 is not reliable in terms of direct matchup wins though and its perhaps 3-4 tiers below the other 3 :bronzong: sets.

:salamence:2 doesn't really want to lead but the lead is somewhat forced and with :Bronzong:1 in the back its more acceptable.
The only possible 3rd comes down to :Honchkrow: vs :Roserade: and there's no doubt that :Roserade: is the better option. One of my soft rules for drafting is "dont be water weak" - and a lot of water types run ice coverage for :salamence: and would just beat :Bronzong:.
The fact that :Honchkrow:4 is also sharing an ice weakness as well as the fact that its just slow (and sucker punch is unreliable) makes the decision easy.
A very realistic alternative was to lead :Roserade: and going over this again I am really not sure if that would have been better or not. While :roserade: is a slightly worse lead, its an easier long-term decision in terms of "fixing" the team and making it stronger.

Playing this team is going to be harder than some may assume, because I will want to position the team around not losing to a random fire type if :Salamence: drops, and also weighing up the best last mon option for :bronzong:.

Team: :salamence:2/ :Bronzong:1/ :Roserade:4

Battle 1: :Magnezone:4 / :Articuno:4 / :Kangaskhan:3
:Magnezone: is a terrible lead to see. :salamence: and :Bronzong: both (very likely) lose to it and :Roserade:'s 2hko would be a roll and accuracy-reliant.
It's important not to overreact in hindsight but this is maybe a point towards leading :Roserade:.
:Magnezone:2 is a special case as because its physical, a small DD setup may be viable. But the risk of staying in against Mag 1 or 4 is unacceptable.
:Magnezone:3 is also a special case as it will 2hko :roserade: on the switch if it uses flash cannon, but it should favor using twave anyway.
On pure odds, the best play is to go :Roserade: blindly.
:Magnezone: reveals tbolt + damage (literally was a max roll from 252 modest) reveals mag4.
This is still "bad" and I did not fully consider the possibility that tbolt + flash cannon could 2hko :Roserade: even with modest but clearly the absolute highest rolls make it possible. Leaf storm is not only 10% to miss, but another 9% from brightpowder and the 2HKO is not guaranteed even if both hit.
That said, the only viable option at this point is to leaf storm twice. Flash cannon does not get a super high roll and :Roserade: does not miss.

next in against a -2 spatt :roserade: is :Articuno:(!) which is also very concerning. :Roserade: is at 2hp so "anything will kill" making the next move hard to predict, although sheer cold is not possible as it is not considered a regular damaging move.
:salamence: is obviously not coming in, the main question is if it's worth saving a 2hp :roserade: for later at the cost of potentially risking a frz, water pulse confusion, etc on :bronzong: (as well as the damage on :bronzong:).
:Roserade: is fast enough that intuitively the answer feels like a "yes". The last pokemon is not steel or flying - and so leaf storm/ sludge bomb are likely to significantly contribute.
Air cutter reveals :Articuno:4 which is definitely the worst set. If sheer cold hits now then the game is likely over. Sheer cold is the most likely move according to decomp analysis - at 62.5% compared to 37.5% ice beam.
Trick Room is an option here as the cost of getting 30% flinch is giving :articuno: a free outspeed on both :Salamence: and :roserade:. TR adds considerable risk as it essentially forces :bronzong: to win the 1v1, but Iron head + fire fang is not guaranteed to KO :articuno: and :salamence:2 + 2hp :roserade: isnt amazing odds for the final 2v1 anyway.
With hindsight I think this choice could be critized. After TR the chance of Sheer cold going off is around 13-14% factoring in all relevant luck (flinch - chance to be chosen - chance to miss). If sheer cold hits next turn, then the :bronzong: decision doesnt matter.
:Articuno:4 goes for Sheer cold on the TR turn but does not hit it.
Even without Charti berry, rock slide would be a mistake. Iron head will cleanly 2hko even without investment.
:Articuno: flinches and the worst of the danger is now over. The next pokemon will come in under TR, but TR will have 2 turns left which is probably the perfect amount as levitate :bronzong: is unlikely to die in a single turn.
:Kangaskhan: comes in and the game is over, as no set will outspeed :salamence: and the endure set is simple to play around via an additional trick room.

Hint: Grass

Grass is the best type hint to see as its by far the least scariest to face.
:Magnezone:4 is a lot better than :Bronzong: in general and the weakness to ground is honestly fine with :salamence: around. Again its a shame that :salamence: is locked into the lead slot.

Team: :salamence:2/ :Magnezone:4/ :Roserade:4

Battle 2: :Politoed:2 / :Exeggutor:3 / :Sceptile:1/3
:Politoed: is an annoying unknown but unlike earlier, the belly drum set is harmless and instead the swagger and hypnosis sets (1/4) are a lot more annoying.
I nearly forget that :Magnezone:4 does not block :Politoed:1 as its lax incense and not brightpowder.
The safest play by far is to go :Roserade: as natural cure covers hypnosis and once it uses swagger, it will not swagger :Salamence: again on the switch back.
even if :roserade: is lost to a crit ice beam, this is "ok" as I know the final 2 are both grass types which are likely to get bullied by :salamence:.
:Politoed:2 is revealed via dive, and this is an excellent setup opportunity for :salamence:. Although Hypnosis is annoying, I can safely wait for it while -2 :politoed: uses dive.
The question of how many DD to take is mostly a question of greed. at +2 :salamence: drops most of the factory with a neutral dragon claw (you'd need ~ 90/90+ bases to live). Grass types aren't really known for this kind of bulk and :salamence: probably forces its way through :Cradily:/:Tangrowth:/etc anyway.
I could go further but trying +3 or more does risk running into a hypnosis + perish song which - while it wont kill :salamence: - does deny the +2 attempt.
Another point in favor of the +2 is that I avoid more random dive chip damage / potential crit. A healthy +2 :salamence: should beat a lot of the things it wont OHKO anyway.
After :politoed: drops to thunder fang, :exeggutor: comes out.
The "risk" of 5% miss on fire fang is worthless here (dclaw is almost identical damage). At 138hp :salamence: should be safe from anything but a crit. I just dclaw twice as set 3 is revealed.
(set 3 was probably the most likely due to its explosion winning most switch in checks).
Problem is, set 3 uses Trick Room. Explosion is now very likely at low HP, and I would rather keep a 138hp :salamence: around for an unknown final grass type.
:Magnezone:4 is both likely to be least useful for the final matchup and also the mon that survives explosion. I switch.
:Exeggutor: does not boom immediately but instead zen headbutts.
there is no real need to stall TR when it probably benefits :Magnezone:, and explosion does enough that i would rather not get hit by it. :Exeggutor: does not explode again (despite explosion being a ~90% chance to be selected)
Last in is :Sceptile:. If there is any risk at all of losing it would be to giving this thing overgrow. After Tri attack set 1/3 are revealed and the match is over (:salamence: always beats them)

Hint: Flying

:Roserade:4 comfortably outclasses both :Sceptile: sets, the team has no real need for speed, and :politoed:2 sucks.
Keeping :roserade: is not ideal for a flying trainer but the other options are worse, both short and long term.

Team: :salamence:2/ :Magnezone:4/ :Roserade:4

Battle 3: (24iv) :Crobat:4 / :Hypno:4 / :Honchkrow:4
24IV trainers are always mixed fortune. I don't have to scout, but set 4 is usually the deadliest set.
On paper, :salamence:2 can setup on :Crobat:4. +1 Thunder fang will always KO (and dclaw probably would) after -1 brave bird recoil.
That said, its a needless risk. What if :crobat: crits, what if i get a 5% miss? It's not like +1 :salamence: locks out the game anyway - i would still be forced out if something like :Articuno:4 came in next.
I instead go to :Magnezone: which risklessly destroys :Crobat:4 while taking barely 13%.
:Hypno:4 comes in which will always lock itself into specs focus blast. I know that the final pokemon is a flying type.
I could sack :Roserade: here and force it to struggle on :salamence:2. In such a scenario I probably end up with a +6 :salamence: at around 70-100hp alongside a 124hp :Magnezone: vs an unknown flying type.
I do not actually think this is the safest outcome compared to trying to preserve :Roserade: (above ~60%HP or so). Any flying set with an accuracy item threatens to beat :salamence: with a single 10% proc.
That may seem like a niche list but it really is not; :Zapdos:4 and :Charizard:4 could both threaten this and in that context 10% miss is quite a lot.
In hindsight a compromise probably works best: which would be to take :salamence: to +2 but not below ~130hp (enough to force :Zapdos:4 or :charizard:4 to get 2 misses).
All this planning is made irrelevant by the arrival of :Honchkrow:4 which is a completely free win.

Hint: -

None of these are viable swaps.

Team: :salamence:2/ :Magnezone:4/ :Roserade:4

Battle 4: :Lickilicky:3 / :Houndoom:3 / :Dragonite: ?
:Lickilicky: is not strong but the variance in how different its sets behave is tricky.
:Lickilicky:1 (belly drum) threatens to heavily punish a switch out, while :salamence: staying in against it is likely to freely create a 3v2.
Losing :salamence: to set 2 (ice beam) would be a huge loss though. It would be a lot better to go :Magnezone: there.
Set 3 (curse) only really matters in the context of playing around lum rest. This should be easy with any play pattern, except keeping :salamence: in after turn 1.
Set 4 is a slight annoyance but can be safely put to -2 attack or lower via :salamence:/ :magnezone: switching.
With a worse (or much stronger) team I probably take the risk and stay in. But with this team and this close to the record I do not want to place myself so far behind with a reckless 75% decision.
I decide that the threat of set 2 putting me behind is a lot worse than what anything else threatens (which is to just basically KO a single one of my pokemon).
:Magnezone: comes in and it reveals curse (set 3). This is fine and :Magnezone: kills it.
:Houndoom: comes in next and as usual its only the special :houndoom: sets (3/4) that are threatening.
Sacking :Roserade: is a real option and the one I ultimately decide on, as :salamence: does not have the bulk to accept any kind of bad luck - and if any bad luck happens to :salamence: then the game could immediately end.
Flamethrower reveals set3 which is the worst case scenario as it runs will o wisp.
Playing into more worst case scenarios, I would prefer to DD in the event of a burn since that at least leaves :Salamence: open to 2hko :houndoom:. I know the final pokemon is not a fire type, and the worst situation then (ground) can at least be forced to eat a dragon claw.
:Houndoom: does not use Will-O-Wisp and :Dragonite: comes in.
Ideally I scout the :dragonite: set, but there's absolutely no way I'd risk potentially losing to :dragonite:4 (which 2hkos a slightly weakened :magnezone: and can survive +0 dragon claw) or set 3 (which can ohko :magnezone: with flamethrower).

Hint: -

Unknown :Dragonite: for :Roserade: is a very real option. I do not mind running 2 4x ice weaknesses at all since this is a statistically unlikely weakness compared to running a bad/ questionable set (like :Roserade:) (this applies even more with :magnezone: as a pokemon that could cover something like :weavile:).
This would also "fix" the lead problem the team has, where I constantly want to rescue :salamence: from unknown turn 1s. With :Dragonite: and :magnezone: in the back I can be a lot more aggressive.
The actual downside to :dragonite: is not the type overlap but the speed. No :Dragonite: sets run any speed investment and while this still makes them strong, it reduces their situational impact in messier games.
In the end I stick with :Roserade:. This was a close call though, and I think presented with the same option again I'd swap. Perhaps the fact that the next game would be WR made me bias more conservative.

Team: :salamence:2/ :Magnezone:4/ :Roserade:4

Battle 5: :Porygon-Z:2 (download) / :Golduck:2 / :Aggron:1
:Porygon-Z: is an uncomfortable lead. Set 3 (ice beam) scares me enough to want to go :magnezone: almost no matter what.
:Magnezone: can lose the 1v1 after the switch to other sets but it should at least knock :Porygon-Z: into kill range for the faster :Salamence:/ :roserade:.
After trick porygon actually dodges tbolt via the "tricked" brightpowder from :Magnezone:.
This rapidly transforms the situation from "fine" to "not ok" as :Porygon-Z: is faster and will 2hko from here.
The "smart" ai comes in and rescues the situation with an Arguably useless thunder wave, which is enough for :Magnezone: to close out the 1v1.
But (!) This is actually a mistake. the clearly better play is to go through :roserade: since :Porygon-Z: always drops to leaf storm and :roserade: is a lot less important.
:Golduck: comes in and low HP PRZ :magnezone: is definitely a sack option.
This works fine as :Golduck: is actually :Salamence:2 setup fodder. I decide to simply boost until :Salamence: enters crit range, and :golduck: scores the crit after only 3 attempts.
:Aggron: comes in and as mentioned, dragon claw is more optimal than either fang option, as if its the sash set then a 5% miss can turn into a loss.
:Aggron:2 here would punish the :roserade: mis-step on turn 3 if QC activates on :Roserade:. To be clear, if i lost to this it would not be bad luck but a direct consequence of slightly suboptimal play.
:Aggron:1 is revealed instead which requires a QC + Crit to beat :roserade: and doesnt get it.

Hint: -

:Porygon-Z:2 is not a good option to take due to how limiting it can be.

Team: :salamence:2/ :Magnezone:4/ :Roserade:4

Battle 6: :Glalie:3/ :Weezing:1/ :Yanmega:2
:Glalie: is a clear :magnezone: scout. As mentioned, physical :glalie: can safely be intimidate stalled to -3.
:Glalie:3 is revealed, which is fine outside of a random freeze. :Magnezone: takes it out.
:Weezing: is an unusual switch in (since its STABs and typical moveset do not hit mag hard) and heavily suggests a fire move (i.e. set 4).
:Weezing:4 would be a small issue due to its speed tie with :magnezone:. There is a chance it is not :weezing:4 however and tbolt has a high chance to pay off.
:Weezing:1 is revealed instead. This guarantees that the final pokemon also has terrible moves into :Magnezone:.
I tri attack as its the least useful move and grudge is likely to strip the PP.
:Yanmega: last is not a surprise and has no realistic shot.

Hint: Ground

Since its the last battle, i could throw Mag out for :weezing:1. This is the kneejerk reaction and somewhat incorrect as baiting ground moves is fine with :Salamence: as a switch, and :Magnezone: actually beats quite a lot of ground types due to their secondary typing. :Magnezone:4 is also likely to be better than :Roserade: against some "generic 3rd mon that isnt ground".
Getting rid of :Roserade: for :weezing: is the better option mostly because of :Gliscor:, :Garchomp: and :Flygon:. These 3 (along with Swampert) are the reason why ground is a scary type spec.
I am scared enough of :Garchomp: to take :weezing:. You may wonder "what would it do" and the answer is that grudge can strip outrage/ dragon claw pp, at which point :Salamence: can return.
For this I do give up the option to leaf storm something like Swampert but I reason that 3/4 swampert sets are physical and can be largely switched around, and Swampert3 is something the team does actually stand a chance against, unlike :Gliscor: and :Garchomp:.
The :Gliscor:/:Garchomp:/:Flygon: matchups still work without :Weezing: since Dragon/steel is a good defensive pivot spam with intimidate. But if they come in later (after one or both of :Salamence:/mag is out of action) then the game could end extremely quickly.

Team: :salamence:2/ :Magnezone:4/ :Weezing:1

Battle 7: :Garchomp:3/ :Torterra:4
:Garchomp: is a terrible lead to see because this :salamence: set, while jolly, does not run outrage.
Still it is a lot better that :Garchomp: appeared earlier in the match while the team is still equipped for it.
:Garchomp:3 would be the worst result although :Garchomp:1 is also bad since it would require a DD to outspeed.
But whatever set it is, :salamence: can't stay in. I go :Magnezone:.
Set 3 (sash) is revealed and my hand is somewhat forced to stay in. Magneonze can flash cannon to bring this into dclaw range and its job is largely done.
:Garchomp: crits aqua tail on the flash cannon which is annoying as it probably takes :magnezone: out of action for later.
I decide that even aqua tail chip of :Salamence: probably isnt worth it and stay in to flash cannon again.
:Torterra: comes out and unfortunately :magnezone: is at such low HP that this could click almost any move, making it too risky to switch out.
:Magnezone: goes down while getting off a signal beam.
:Torterra: is not low enough for me to have faith in dclaw/fire fang and i'd rather :salamence: not get hit by stone edge (this can happen via QC anyway) and i'd also rather preserve the intimidate as an option against the final mon.
:Weezing: comes in and finishes it off with sludge bomb.
:Pinsir: comes out and is unlikely to win. Notably sludge bomb + fire fang is probably guaranteed, so there is no need to fall back on shock wave insurance against lax incense. :Weezing: just directly wins this 1v1.

As a brief postmortem, :Roserade: would also have been fine here with intimidate assistance. I don't regret the :weezing: swap, but its one of the closer calls in the run.
At the time of writing, this writeup has now exceeded 16,000 words and a lot of hours; for the sake of ever completing it I will now stop such detailed battle breakdowns.
The videos show my live thought process throughout the drafts and battles either via text or commentary. While neither is perfect (I am aware of audio issues) it does shred some context.
Instead I will largely focus on drafts, mistakes and any particularly weird turns.
(2:37:00 in)

draft: :Abomasnow:3/ :Electrode:4/ :salamence:2/ :Lopunny:1 (Klutz)/ :Walrein:2/ :Raikou:2
hint: -

:abomasnow:3 and :electrode:4 are both slightly above average sets although they dont work terribly well together (sash/snow)
:salamence:2 returns and its a decent pick.
:Lopunny:1 and :Walrein:2 are both below average but not absolutely terrible.
:Raikou:2 is the only bad :raikou: set and is also below average.

I am a fan of :electrode:4 due to how often it "works" and contributes more than just a single trade. That said it definitely has its flaws and falls short of being a "real" set outside its gimmicky moveset.
I could lead with :electrode:4 (for a single battle it would be the best decision) but I'd rather not commit to that on battle 1 of 7 as its likely to be exchanged and I will have forced lead blindness if swap options are limited.

I am once again stuck with a :salamence:2 lead and decide to just take the 2 individually strong mons (:abomasnow:3 and :electrode:4). this is a draft i expect to change.

Team: :salamence:2/ :Abomasnow:3/ :Electrode:4

Battle 1: :Drapion:1 / :Dugtrio: ? / :Infernape: 2
The only interesting decision here was to sack :abomasnow: for a 2nd intimidate on :infernape:.

I ditch the shaky and limiting :electrode: for :Infernape:2. :Infernape:2 is the worst :infernape: set but works better with :Salamence:/aboma.

Team: :salamence:2/ :Abomasnow:3/ :Infernape:2

Battle 2: (24iv) :Porygon-Z:4 (Adaptability), :Gardevoir:4 (sync), :Breloom:4
The biggest mistake i made here was missing the trainer name.

Adaptability :Porygon-Z:4 is a huge upgrade over :Abomasnow: and I think most players would snap pick this.
However, I actually prefer the often underrated :Gardevoir:4 with the intimidate lead that's already looking to pivot.
:Abomasnow: and :infernape: are of a similar approx. power level, but I'd rather keep :Abomasnow: as I hate being too water-weak in a draft.

Team: :salamence:2/ :Abomasnow:3/ :Gardevoir:4

Battle 3: :Armaldo:4/ :Claydol:1/ :Moltres:2
The ending of this match was dicey under TR as the AI rolled the best :moltres: answer to the situation.

:Moltres:2 is actually a viable swap despite my previous comments on waters. This is simply because with Sunny day it flips enough of these matchups, and :Moltres: just has good numbers.
I do not like my draft at this point but I would call the team effective.

Team: :salamence:2/ :Moltres:2/ :Gardevoir:4

Battle 4: :Jynx: ? / :Steelix:1/ :Houndoom:1
This is a really simple win

No swap as all 3 of these would introduce more weaknesses and issues than they could solve.

Team: :salamence:2/ :Moltres:2/ :Gardevoir:4

Battle 5: :Machamp:1 / :Kangaskhan:4/ :Hippowdon:4
I once again feel the need to safely evacuate :Salamence: from the lead. Once again this is the "problem" with lead :Salamence:2.
:Gardevoir:4 proves itself here by not only pivoting in but taking 2 pokemon out while doing so, creating the 2v1.
Hippo does get a free shot at hitting a QC Stone edge to end the run but misses it.
If this had happened I would have blamed the overly safe turn 1 decision. The 16% loss would have been my fault, not bad luck.

next hint is grass which I'm already overspec'd for.

Team: :salamence:2/ :Moltres:2/ :Gardevoir:4

Battle 6: :Abomasnow:2 / :Pinsir:3/ :Ludicolo:2
Learning the lesson from the previous battle i decide to not coddle :Salamence: and do just click the obvious 4x effective move.
Ending was important to optimise via Synchronise. Being dumber than this could have lost the game.

Team: :salamence:2/ :Moltres:2/ :Gardevoir:4

Battle 7: Mismagius1/ :Floatzel:2/ :Scizor: ?
Nothing noteworthy here beyond the slight optimisation for :Floatzel: paying off.
At this point as a meta-criticism i think continuing IRL was a mistake. Almost 4 hours in I am not as fresh for constant analysis and decision making. It would have been better to stop (I continue this session for another 3 hours).
My justification was that I was expecting the run to be over already/ soon.
I do make a massive mistake in the next round (see battle 3) and I think fatigue is partly to blame.
(3:52:00 in)

Draft: :Mr. Mime:3 (soundproof)/ :Scizor:2 (Swarm)/ :Electivire:4/ :Glaceon:4/ :Vaporeon:3/ :Hariyama:3 (Guts)

:Mr. Mime:3 is bad.
:Scizor:2 is good but something of a gimmick.
:Electivire:4 is a very good set and also a functional lead (lucky, because nothing else is).
:Glaceon:4 is a solidly good-ok set.
:Vaporeon:3 is a good set on average but the low acc moves are something I'd rather avoid if possible.
:Hariyama:3 is definitely below average and without Thick Fat its perhaps even a bit worse than that.

The first hint is Psychic. :Electivire:4 is an easy lead. I then just go with :Scizor: and :Glaceon: as backups.
This is a flawed draft in that the weakness to fire is real but I am likely to be safe for a single match.
A close alternative here would be :Vaporeon: over :Glaceon: and there is logic behind this in :Vaporeon: being arguably better and also better on average into psychic types which are more often specially based.
I am at least content that :Scizor: covers some of the nastiest options such as Lati@s, :starmie:4, etc.

Team: :Electivire:4/ :Scizor:2/ :Glaceon:4

Battle 1: :Bronzong:4 / :Medicham:4/ :Ursaring:1
The :scizor: vs :Bronzong: 1v1 might look odd but I am simply playing it in a way to ensure I don't lose to crit (unless immediately followed by a QC). The PP on roost is almost certainly irrelevant compared to how bad losing :scizor: would be.
Conversely against :medicham: its never worth risking the crit to attempt to heal very gradually.

:Bronzong:4 is a pretty good set and certainly beats the Gimmicky :Scizor:2. Even if its not levitate thats ok.

Team: :Electivire:4/ :Bronzong:4 (Heatproof)/ :Glaceon:4

Battle 2: :Lucario: ? / :Regirock:1 / :Flareon:3
This more than anything highlights the power of going for a messy/ damaged 3v1 rather than a cleaner 2v1.
The 3v1 line is particularly good as :bronzong:4 runs explosion, which means there is always an option to sack and immediately go to detonate. This makes most AI comebacks extremely rare, but its a common Factory motif.

I will take an unknown :Lucario: over :Glaceon:4. The only outcome i'd be upset with is :Lucario:2 and even that is basically still an upgrade.
The triple EQ weakness is both somewhat concerning and somewhat fake. The AI has a chance to guess :bronzong: has levitate errorneously. :Electivire: has a shuca and both :electivire: and :Lucario: are "fast".

Team: :Electivire:4/ :Bronzong:4/ :Lucario:4

Battle 3: :Poliwrath:1/ :Zapdos:4/ :Gallade:3
:Zapdos:4 is a terrible thing to see for this team the second that :Electivire: takes any damage.
This is the single biggest battle misplay I make. Perhaps I can blame fatigue, but the tl;dr is that I simply forget that explosion halves defense this gen while scoping out possible plays. Academically speaking I "know" this. But I am actively thinking more about brightpowder and how to best simply forget to account for it in the mental calc.
Because of this, in my head explosion is doing ~90% to :Zapdos:4 and is not a safe OHKO. This causes me to avoid making the correct decision (simply immediately exploding) because I assume that iron head is completely necessary as pre-chip.
What is stupid about this is that I would have avoided this mistake if I loaded up any of the frontier battle calculators
As is, Brightpowder does not punish the useless iron head turn.
I make a big deal of :Gallade: live but actually its incredibly obvious that it isn't :gallade:4 since :Poliwrath: CC trivially outdamages :poliwrath: heat wave, so :gallade:4 would have come in 2nd.
set 1 is also reduced in chance as an option because of brick break on switch-in rules.
Set 3 is the only set :lucario: beat but I made a bigger deal of this at the time than was necessary. The actual luck was more like a ~60% outcome.
This is still one of the absolute scariest games (and would have been even without the EQ mistake). At the same time I couldn't have played the start any different; it would have been stupid to over-protect :electivire:4 against an unknown :poliwrath:.

:Zapdos:4 is easily the best take and one of the absolute best factory mons but its an annoying switch. I am uncomfortable doubling down on electric as a double STAB coverage move on a team of 3 where the final member would either be heatproof :bronzong: or :lucario: (aka ground is a concern, even with :zapdos:)
I am forced to give up :Electivire:4 (objectively speaking the best current team member) for it.

Team: :Zapdos:4/ :Bronzong:4/ :Lucario:4

Battle 4: :Shiftry:4/ :Staraptor:4/ :Mr. Mime:4
Maybe the easiest battle of the run.

:Staraptor:4 is an incredibly strong set and I happily leave :lucario: behind for it. It functions best as a lead but :Zapdos:4 trumps it and :Zapdos:4 is probably never getting replaced.

Team: :Zapdos:4/ :Bronzong:4/ :Staraptor:4

Battle 5: :Meganium:2/ :Heracross:4/ :Skuntank:4
Another easy battle

:Heracross:4 while strong is also unreliable and there's no reason to accept that.

Team: :Zapdos:4/ :Bronzong:4/ :Staraptor:4

Battle 6: :Golduck:4/ :Regice:2/ :Salamence:4
I once again get stuck in the wrong gen mode for explosion. Luckily this time it doesnt really matter.
:Salamence:4 is scary but its on the list of things :Staraptor:4 handles.

:salamence:4 is one of the absolute best sets in the factory. It also makes a lot of sense to swap in for Fire.
I do not want to go full-flying as there are enough sets like :Aerodactyl:4 or fast electrics to expose that. Instead :Staraptor: leaves.

Team: :Zapdos:4/ :Bronzong:4/ :salamence:4

Battle 7: :Weavile:4/ :Typhlosion: ? / :Heatran:1
Minimal resistance here. I get some unnecessarily good luck with the QC but its not like it mattered.
(5:11:00 in)

draft: :Drapion:1/ :Luxray:1/ :Weavile:1/ :Politoed:2/ :Articuno:4/ :Gardevoir:3 (trace)

:Drapion: and :Weavile: are both good to ok.
:Luxray: is not good.
:Politoed:2 is definitely bad.
:Articuno:4 is good but not great.
:Gardevoir:3 is slightly better than :Articuno:.

There is not a whole lot of thinking to do here, we take the good pokemon. Only 1 of :Weavile: and :Drapion: can be on the team since they have significant overlap.
:Weavile: gets left behind since I slightly prefer :drapion: in terms of reliability and I also like how it functions as a lead which often baits out Earthquake.

Team: :Drapion:1/ :Articuno:4/ :Gardevoir:3

Battle1: (24iv) :Gastrodon:4 / :Aggron:4 / :Machamp:4
Not a lot to say here as this is a very normal factory win where my sets are slightly better and I'm able to enter into favorable matchups.

:Aggron:4 is actually a good lead and a good example of a lead that functions despite a speed issue. This is because it is rare for it to fail to work via either direct attacks or metal burst.
I definitely do not recommend :Aggron:4 in general. Outside of the lead slot its just "ok".

Team: :Aggron:4/ :Articuno:4/ :Gardevoir:3

Battle2: :Miltank:2/ :Vespiquen:2/ :Regice: ?
Very rare situation where going for a single sheer cold is correct, and its specifically because of how useless the chip damage was otherwise.

A few swaps are a bit tempting but I agree with the decision to stick in hindsight.

Team: :Aggron:4/ :Articuno:4/ :Gardevoir:3

Battle3: :Gliscor:2 / :Luxray:1/ :Umbreon:1
There is a small trap in this battle, which is how badly Trace rivalry could have gone wrong.

Although Gard3 is good, :Gliscor:2 is better. You can largely ignore the fang moves, focus on the part where this is a fast 252/252 jolly set with STAB EQ, immunity to EQ and an impressive base def.

Team: :Aggron:4/ :Articuno:4/ :Gliscor:2

Battle4: :Tauros:4/ :Manectric: ? / :Ambipom:1
The way :Gliscor: just completely dominates some sets is funny.

:Tauros:4 is an incredible set in the factory. :Articuno: actually leaves before :Aggron:4, as this team does not struggle to close out 2v2s and :Articuno:4 eternally threatens to be useless.
This said, taking a lead :Tauros: for the longer run would have been fine.

Team: :Aggron:4/ :Tauros:4/ :Gliscor:2

Battle5: :Ninetales:4/ :slowbro:3/ :Drifblim:3
:Ninetales: immediately turning up in the lead after i decide against leading :Tauros: almost feels intentional from the factory.
:Slowbro:3 is a fairly large punish follow up, and :Tauros: misses its roll but so does :Slowbro:.
The ending to this match is fairly crazy. I do think that based on all available information, sacking :Gliscor: is probably correct but relying on Stone edge was concerning.

:Aggron: vs :slowbro: lead is actually close but the chance of that :Ninetales: being a hypnosis set has spooked me away from the already known hit-by-status inconsistency of :Aggron:4.

Team: :Slowbro:3/ :Tauros:4/ :Gliscor:2

Battle6: :Porygon2:2/ :Manectric: ?/ :Mamoswine: ?
A relatively straightforward battle

No swap makes sense.

Team: :Slowbro:3/ :Tauros:4/ :Gliscor:2

Battle7: :Golduck:4 / :Gengar:2/ :Ambipom:1
Another PP stall festival against a hypnosis set. Although I am grateful this was not the Psychic set.

This is an unusual full swap round where nothing in the initial draft carried through to the end.
From here on there is live commentary (with a scuffed mic) in the video which reduces the amount of text in the video but I'm continuing with a more limited summary.
Some of my comments make no sense without the context of twitch chat but it's not like i'm giving any kind of genius insight while mostly focusing on the battles anyway.
If you want watch a bit of the run and get some insights into what I'm focused on while playing then i recommend round 15+ (all rounds from here on have live comm).
(0:40 in)

draft: :skarmory:4/ :Weavile:3/ :Hypno:1/ :Flygon:2/ :Absol:3/ :Jynx:2
Hint: grass

This draft is terrible and I am fortunate to see the grass hint as grass is one of the absolute worst types in factory - particularly when piloted by the AI.

:Hypno:1 and :Flygon:2 are absolutely terrible. :Hypno:1 is far worse as :Flygon: can at least function as a fast cleanup.
:Jynx:2 is a pokemon I would never realistically ever want to take but in a draft as bad as this its probably on the table.
:skarmory:4 is a worse set than many may assume although I do still classify it as somewhat average.
:Weavile:3 actually is ok with its sash, speed stat and counter although the lack of ice STAB does feel horrendous
:Absol:3 is definitely a bad set but it is not in the same tier as something as bad as :Hypno:.

As mentioned in live comm, I do think the heads up play is to focus on the grass spec for the :jynx:/:Skarmory:/:weavile:/:absol: shortlist since its likely that I can swap the worst or least synergistic option.
Grass spec also introduces hypno1 slightly back into the equation due to how how psycho cut does overlap with a lot of their secondary types.
:Weavile:/ :skarmory: are pretty much locked in. The 3rd could be anything except :Flygon:. I pick :Absol: since a lot of its shortcomings (not quite fast enough, not quite bulky enough) dont apply to a grass opponent and it does run x scissor.

This is part of my guide on the factory but this draft is bad enough that I actually want to intentionally lead with the worst set (:absol:) as its the most likely to be swapped out and i can scout for not only swaps but lead swaps.
The downside is that this usually makes the first match a bit riskier. But as mentioned above I am basically specialising for grass anyway and if :Skarmory:4 can do one thing its switch into and then beat grass types.

Team: :Absol:3/ :skarmory:4/ :Weavile:3

Battle 1: :Exeggutor:/ :victreebel:3/ :Marowak:3
Grass puts up basically 0 resistance here.

To give some context to the commentary; chat wanted :marowak:.
I'd probably have stuck with taking an unknown :Exeggutor: set over :absol: fairly quickly.
In the end I am swayed enough to taking :Marowak:3 but I would underline that Wak3 as a lead has severe issues and there are a lot of lead matchups (probably most water types) that could immediately end the streak.
The most optimal switch is difficult to figure out but I do know there is no good answer.

Team: :Marowak:3/ :skarmory:4/ :Weavile:3

Battle 2: :Rampardos:1/ :Glalie:1/ :Torterra:4
The live comm does cover this match pretty well.

As mentioned, a terrible opponent is both a blessing and a curse because while being a "free" win, it gives no good swap.

Since the time of playing, my opinion on :torterra:4 has gone up massively and I'd actually classify it under the "definitely good" tier. I would not hestiate to swap Wak3 for :Torterra:4.
:Torterra:4 is better than Wak3 in almost every way that matters minus the inconvenience of the grass typing. but in this team, i actually dont mind it as it potentially saves a few water matchups (but definitely not all due to ice beam).
The fact that the team triples down into a fire weakness bothers me a lot less than how weak the team currently is. Being weak to whatever % fire type is is better than the current % chance the team just has to lose into random pokemon.
A lesser weakness is "3 physical mons" but I do agree with old me that something like Cress can rip through the team.
If I could magically adjust the order i would put :torterra:4 in the back but I do not have this kinda of teambuilding luxury.

Team: :Torterra:4/ :skarmory:4/ :Weavile:3

Battle 3: :Leafeon:2 / :Froslass:3 / :Golem:2
I discuss it in live comm but the major decision point here was deciding between a (damaged) 3v1 or a 2v1 with focus sash intact on :weavile:.
There are enough examples on both but my bias is definitely towards preserving sash since :Skarmory: isnt that fast and since I feel as if the list of special sets that go through :weavile: and :torterra: is smaller.

To explain another oddity; the 4x bonus the AI gives moves that are "double super effective" makes superpower a possible click for once.

:Froslass: is a better pokemon than :weavile:.
:Weavile: is better than :skarmory: but taking 2 ice types in and already type-compromised team is dubious.
:Froslass:3 is also far better in the lead slot although given the current context i do not want to give up :torterra:.
The "fire weak" curse continues if i take it but it does have to be a fairly specific fire type to get through :torterra:4 (for example, :Magmortar:3 can't do that).
This mostly comes down to wanting to fix the lead (get rid of :torterra:) vs wanting to fix the draft (a bit more) (get rid of :weavile:).

I decide that with 4 games left there's considerable utility in fixing the lead in able to then later fix the draft.

Team: :Froslass:3/ :skarmory:4/ :Weavile:3

Battle 4: (24iv) :Porygon2:4/ :Dusknoir:4/ :Garchomp:4
This match was going peacefully until :Garchomp:4 entered in a "1.5v1" situation.
The decision to go :weavile: after dbond felt logical and it does feel like a kick in the teeth when the AI randomly gets a lucky refutation to a more considered play (if i had gone :Skarmory: first then :weavile: easily wins via sash counter).
:Skarmory: still handles a weakened chomp if no burn and no crit though, and the AI does not find the burn. I am scared enough to properly calcaulate LO BB.

:Garchomp:4 is the factory god for a reason. 24IV is a larger nerf than you may expect as :garchomp: loses its magical 102 tier and instead is effectively "98".
I'd normally want to lead but :froslass:3 is easily the best pokemon on the current team and :Garchomp: is of course completely functional in the back.
:Weavile: leaves here in favor of having both Dragon/steel and ground/flying in the back as pivots.

Team: :Froslass:3/ :skarmory:4/ :Garchomp:4

Battle 5: :Cresselia:3/ :Steelix:4/ :Donphan:3
This match is almost scary as :Cresselia:3 is one of very few pokemon that :froslass: cant escape from, creating a 2v1 with a ground type spec that could have gone wrong.

:Cresselia:3 is way way better than :Skarmory:4.
We've now completely got rid of the initial draft.

Team: :Froslass:3/ :Garchomp:4/ :Cresselia:3

Battle 6: :Glalie:3/ :Forretress:2/ :Nidoking: ?
At this point the team is difficult to improve without making it temporarily worse.

Team: :Froslass:3/ :Garchomp:4/ :Cresselia:3

Battle 7: :Starmie:1/ :Tangrowth:4/ :Abomasnow:2
This is a very easy battle.
(2:07:00 in)

Draft: :salamence:4/ :Pinsir:1/ :Nidoqueen:2/ :Vespiquen:2/ :Claydol:3/ :Tangrowth:2

This is a draft that picks itself;
:salamence:4 is one of the best sets and functions very well as a lead.
:Vespiquen:2 and :Tangrowth:2 are both terrible.
:Pinsir:1 is good.
:Claydol:3 is average.
:Nidoqueen:2 is bad.

The 3 good pokemon here actually have decent synergy which is a bonus.
This team can definitely be improved but its an order of magnitude better than 15 as a draft.
Although i wont go into detail here I still do talk through everything that could go wrong in the video.

Team: :salamence:4/ :Pinsir:1/ :Claydol:3

Battle1: :Lanturn:3/:Crobat:2/ :Meganium:3
This was as easy as it looks.

Team: :salamence:4/ :Pinsir:1/ :Claydol:3

Battle2: :Charizard: ? / :Vaporeon: ? / :Machamp:2
This match became easy after :Charizard: forgot to get lucky on turn1.

Team: :salamence:4/ :Pinsir:1/ :Claydol:3

Battle3: :Lucario: ? / :Jolteon:4/ :Tangrowth:4

Although :Lucario: is an unknown set I will take it over :Claydol:3.

Team: :salamence:4/ :Pinsir:1/ :Lucario:1

Battle4: :Gallade: ? / :Tauros:4/ :Steelix: ?

:Tauros:4 is a huge pickup and a straight upgrade over :pinsir:.
This now draft is now "very strong" and should be very resistant to losing to any single set from an opponent.
I'm also able to play :Salamence: even more aggressively if needed which is where it really shines as a set.

Team: :salamence:4/ :Tauros:4/ :Lucario:1

Battle 5: :probopass: ? / :Cradily:4/ :jolteon: ?
When played carefully this offers no resistance either.

Team: :salamence:4/ :Tauros:4/ :Lucario:1

Battle 6: (24iv) :granbull:4/ :Hippowdon:4/ :Yanmega:4

Final is a grass spec and as mentioned in the video I think :tauros: still beats :Yanmega:4 as anti-grass.

Team: :salamence:4/ :Tauros:4/ :Lucario:1

Battle 7: :Bronzong:3/ :Abomasnow:2/ :Vileplume:2
This is actually a very deadly and close battle and a single misplay could have lost here.
:salamence: cannot safely 1v1 an unknown :bronzong: without risking levitate. :Lucario: faints.
:Abomasnow: comes back in after :salamence: drops and obviously forces it out.
Sheer cold then hits :Tauros: on the switch.
And then :salamence: gets leech seeded by :Abomasnow: before :vileplume: comes out. :Vileplume:2 (opposite gender) seriously threatens to stall :Salamence: for long enough for hail + leech to KO it.
:salamence: critically misses the first zen headbutt here and with ~57% HP remaining vs 100% the game is suddenly close.
What follows is a very precise set of calcultions to avoid a loss to potential confusion while trying to balance the miss chance on zen headbutt.
The AI slightly misplays (not in a way that would have changed the outcome given the luck in the game, but in a way that could have mattered with even worse luck).

This battle serves as something of a warning for what the factory can randomly throw at the player. I'm not sure what line would have been better though as post-49 I am forced to scout these sets.
Perhaps :salamence: is supposed to just crunch :bronzong: and hope to not get 2hko or hope for no TR or whatever. With hindsight obviously :lucario: + :tauros: easily handle :abomasnow: + :vileplume: but knowing these 2 were grass types as the only information, i felt like it was a lot safer to keep :Salamence: around in case of :Torterra:/ :Exeggutor:/ :Leafeon:/ :Breloom:/ etc.
An exhaustive analysis of every possible grass duo that comes out later may perhaps reveal what the better line was, but I suspect saving :salamence: was correct
 
(cont..)
(3:06:00 in)

draft: :Wailord:1/ :Golduck:1/ :Breloom:1/ :Electrode:3/ :Lucario:1/ :Lopunny:1 (Klutz)
hint: -

Both :lopunny: and :electrode: are bad and not in contention.
:Lucario: is the best set here and also functions well as a lead due to its speed and specific type.
:Breloom: is also good although not in the same tier.
:Wailord: vs :Golduck: is weighted towards :golduck: for sure. Faster, confuse ray is something of an insurance, rindo is potentially relevant and it also just hits slightly harder.
The draft would not be that different with :Wailord: but I consider :Golduck:'s edge enough to classify taking it here as a mistake.

Taking :Breloom:1 in the back is a bit limiting as a lot of its utility is locked into the focus sash item meaning it cannot often come in. It's probably the most important thing to swap out.

Team: :Lucario:1/ :Breloom:1/ :Golduck:1

Battle1: :Exploud: ? / :Donphan: ? / :Gyarados:1
Pretty simple win.

The edge :Gyarados:1 has over :Golduck: is small but relevant and with a lead :lucario: it becomes quite large. I would again say no swap would be a mistake.

Team: :Lucario:1/ :Breloom:1/ :Gyarados:1

Battle2: (24iv) :Typhlosion:4/ :Claydol:4/ :Honchkrow:4
This is a pretty bad team for a set4 only trainer.

:Claydol:4 is probably the best swap. In a vacuum :Breloom: is a slightly better set but :claydol:4 works better as a bench mon with the :lucario:/:gyarados: teammates.
I also consider :Lucario:/:Gyarados: to be relatively safe in a 2v2 so :Claydol: having explosion as an option is more useful than it would be on a worse team.

Team: :Lucario:1/ :Claydol:4/ :Gyarados:1

Battle3: :Hippowdon:2 / :Snorlax:4/ :Floatzel:2
:Snorlax: proved something of a problem but I feel that the pivoting into an eventual :claydol: was the best play to keep options open for later, given that hippo seemingly cant work against :gyarados:.
What follows is yet another PP stall fest that gives a small reward of a rain-prepped :gyarados: into the final 3v1.
There is no payoff though as its a water type.
I am almost caught out by the Focus punch here as i initially assume its :floatzel:4 because of the previous switch in.
What follows is a precarious 3v1 where :floatzel: obviously outspeeds all 3 and also gets a crucial DT on the switch out i do to repeat intimidate.
I then also _maybe_ (?) misplay the penultimate turn. Aqua tail will slightly outdamage bite (at the cost of 5% acc) although I was surprised that :floatzel: wasnt in range of bite after EQ. Another case where I should probably have done a "full" calc instead of relying on eyeballing HP bars and going by feeling.
The first bite miss was scary.

:Floatzel:2 could have badly punished the team in this match under a different set of choices. For example, keeping :Gyarados: in on :snorlax: instead of going for the intimidate switches would have been a disaster.
This is the kind of small optimisation which likely saved the match for once. Even with how slowly and precisely I played the match, :Floatzel: still almost won.
At the same time, perhaps "misplaying" earlier would have improved the :floatzel: matchup since it would have struggled to 1v3 in the sand. but again its impossible to know such things and setting up rain for :gyarados: definitely felt correct.

:Snorlax: and Hippow (even though its set 2) are viable takes although I do prefer :Snorlax:. :Claydol: makes way.

Team: :Lucario:1/ :Snorlax:4/ :Gyarados:1

Battle 4: :Cradily:2/ :Jynx: ? / :Zapdos:1
I make a small mistake in the :jynx: set identify here.
It definitely was not :jynx:4 as that is the one that would outspeed and the one i knew :gyarados: could come in safely against.

Taking :Zapdos:1 is very tempting and although its standalone better than :Snorlax:4 it doesnt really fit on this team as it opens up a weakness to electric as well as shutting off an effective counter to psychic.
:Zapdos:1 also doesnt function as a reliable counter and switch in due to the unreliability of thunder. The "synergy" with rain dance :Gyarados:1 is somewhat fake as in difficult games I may never have time to rain dance anyway.
:Lucario: is probably a better lead than :Zapdos: so I dont think thats a logical option.

Team: :Lucario:1/ :Snorlax:4/ :Gyarados:1

Battle 5: Mismagius2/ :Togekiss:3/ :Miltank:3
Pretty simple match as mismagius2 is terrible.

In hindsight/ with more experience I disagree with my swap decision here.
A better decision would have been to swap out :Gyarados:1 for :Togekiss:3. This is because not only is :Togekiss:3 significantly better but also because it has less variance. Even if it is Hustle I would make the swap.
Not doing this is inaccurate from me and enough to be called a mistake. If I lose in the next 2 games I can probably point to this as the cause.

Team: :Lucario:1/ :Snorlax:4/ :Gyarados:1

Battle 6: Regigigas1/ :Granbull:1/ :Shiftry:1
straightforward game against bad sets.

Team: :Lucario:1/ :Snorlax:4/ :Gyarados:1

Battle 7: :Blissey:2/ :Gallade:1/ :Lanturn:2
Rare use for roar here as :Blissey: would have stalled all the aura sphere PP (spamming recovery is something the AI does do consistently and its a 4hko).
using Double edge or EQ on :blissey: could have going horribly wrong given what was in the back.
Allowing :Gyarados: to be hit by toxic could also have gone badly wrong.

:Togekiss: would have made this match a lot more straightforward as after flash cannon I'm sure air slash KOs :gallade:.
(2 mins in)

draft: :Lanturn:4 (Volt absorb)/ :Typhlosion:1/ :Magmortar:4/ :Gastrodon:2/ :Regirock:2/ :Gallade:2
hint: -

:Lanturn:4 is a great set, particularly with volt absorb.
:Typhlosion:1 is a bad set
:Magmortar:4 is good by the standards of fire types but not reliable.
:Gastrodon:2 is ok. I perhaps undersell it slightly in commentary.
:Regirock:2 is actually ok although it also has significant variance.
:Gallade:2 is surprisingly decent.

Ideally I don't touch :Regirock: here as the draft is functional enough to avoid such a set.

:Gastrodon:/ :typhlosion: realistically are out of the running since they're outclassed by :lanturn: and :magmortar:.

As explained in commentary, the temptation here is to definitely lead :magmortar: as its a lead i would want to swap out but also functions ok in neutral matchups.
The issue with the :magmortar: lead is definitely that this draft has absolutely no EQ switch-ins.

The alternative (:lanturn:/ :gallade:) have similar problems as leads on this team which lacks any kind of defensive pivoting.
I really do not want to have to ditch :lanturn: as its the best mon in the draft and I want to avoid any kind of lead set swap conflicts.
:Gallade: is another set i want to swap out ideally but i will have to react to what is offered to see if it goes before or after :magmortar:.

Team: :Magmortar:4/ :Lanturn:4/ :Gallade:2

Battle1: (24iv) :Blissey:4/ :jolteon:4/ :dewgong:4
this should be an easy opponent to beat but its visually obvious how limiting the draft is in terms of being "locked" into matchups without easily being able to switch out.

Since the next hitn is grass I'm not going to take anything.

Team: :Magmortar:4/ :Lanturn:4/ :Gallade:2

Battle2: (24iv) :Weavile:4/ :Ludicolo:4 (rain dish)/ :shiftry:4
Not much to say about this.

:Weavile:4 and :Ludicolo:4 are pretty good and roughly as strong although :Weavile: is probably better.
the water hint definitely pushes towards :ludicolo: but the better long-term decision for synergy would be to have :weavile:4 as a much faster bench mon.
This is a bit awkward as it blocks :gallade:2 from being switched out now for speed coverage reasons.
I think the "water" hint is one of the scarier ones though and :Magmortar:4 isnt something that I want to keep.

Team: :Ludicolo:4/ :Lanturn:4/ :Gallade:2

Battle3: :Mr. Mime:2 / :Feraligatr:3 / :Starmie:2 (illuminate)
Lead screens sets are annoying to handle and PP stalling is necessary not just to avoid any hypnosis loss but also to tick down screens enough.
I did slightly overswitch and it definitely did matter very slightly. I should have taken the situation even more slowly.

:Starmie:2 is a frustrating pokemon as its average power is very good even accounting for accuracy.
However "at its worst" it is _very_ bad.
I do err on the side of picking pokemon like this though, or i at least prefer it over having bad sets that never have a shot.
:Ludicolo: has to leave as there is way too much role overlap and I would rather the luck-set be sent out first so i can adjust to how much risk i may need later on.

Team: :Starmie:2/ :Lanturn:4/ :Gallade:2

Battle 4: :Politoed:4/ :ampharos:3/ :Heracross:4
Once again :Politoed: strikes as a mon that is difficult to play around all possible sets.
The best decision after a lot of thought is actually :Gallade:2 I feel, which isnt intuitive but does offer the best average outcome across possible :politoed: sets.
:Gallade: vs :Politoed:4 is interesting though because i both want to remove wacan and get low enough so that ice beam/ psychic KO but I also definitely dont want to lose :gallade: before a few attacks. Hypnosis hitting complicates this even more.
:Gallade: ends up at 52hp which is a very awkward number as its just outside ice beam/ psychic range but just below the range i'd need to somewhat safely escape hydro with via drain punch healing.
If :lanturn:4 wasnt slower then thunder punch would just be correct no matter what.
The way :Gallade:2 constantly baits the AI in this match is a hidden reason for why its pretty good and maybe better than people may expect.

Another small but not important mistake; i completely hallucinate the :ampharos: 3/4 difference in this game.

:Heracross:4 is another "great but unreliable" option although endure/reversal is decently reliable.

Team: :Starmie:2/ :Lanturn:4/ :Heracross:4

Battle 5: :Hippowdon:1 / :Blaziken:4/ :Forretress:1
:Hippowdon: makes :Heracross: very unreliable in this fight immediately.
:Starmie:2 does have a Hydro miss moment here but manages to get the second and not go into the 4% double miss despair.

A double miss here wouldnt have been a loss since :blaziken:4 would not have used superpower (even if it would KO) on :lanturn:, which means :lanturn: 1v3s the entire team.

I could bail out of one of the slightly more RNG sets (:starmie:/ :heracross:) here for the weaker but reliable hippo/:blaziken:.
However hippo's sand stream interferes with :heracross: (which is the one i'd want to keep of those 2)
and :blaziken: is a frustratingly not-fast-enough fire type which is a known issue in extended drafts and also basically the same mon i ditched earlier (:Magmortar:).

Team: :Starmie:2/ :Lanturn:4/ :Heracross:4

Battle 6: :Meganium:3 / :Mr. Mime: ? / :Raikou: ?
:Meganium:3 is legitimately one of the best leads the AI can roll because dealing with dual screens is a pain and stalling them often isnt viable as too many things get smacked with leaf storm.
Light screen does go up and because of that i wouldn't say that the 3v2 is necessarily a lead as if :heracross: drops (particularly drops unexpectedly without endure) then it could end very quickly.
:Heracross: embarrasses :Mr. Mime: with resisted reversal getting the KO. This incidentally is the exact problem with the player drafting these frail psychic types.
I badly want to scout :Raikou: but there is no safe way of doing so (10% thunderbolt prz anyone? :raikou:4 can 1v1 :lanturn:4 also )

The last hint is rock which is a nail in the coffin of the blind :Raikou: swap.
In an earlier battle I would roll the dice on :raikou:. On the final battle into a rock specialist I like sticking with 3 pokemon good against rock types.
On top of this :Raikou:2 is a bad outcome and the current team should be good enough in an isolated match.

Team: :Starmie:2/ :Lanturn:4/ :Heracross:4

Battle 7: :Mamoswine: ? / :Armaldo: ? / :Cradily:4
Even playing out the :cradily: matchup safely (no blizzard) :starmie: gets the 1v3 here
(1:41:00 in)

draft: :Golem:2/ :Mamoswine:1/ :Lapras:1 (shell armor)/ :Registeel:3/ :Dewgong:4/ :Zapdos:2
hint: -

:golem:2: terrible
:mamoswine:1: surprisingly great
:Lapras:1: good
:Registeel:3: ok
:Dewgong:4: ok/bit bad
:Zapdos:2: good

The team that is asking to be built here is :Zapdos:/:Lapras:/:Mamoswine: and it doesnt have any huge weaknesses.
While the backline both being ice is ehhh, the only shared weakeness is fighting and: 1) I dont intend to recklessly sack :zapdos: and 2) EQ + Aqua tail covers a surprisingly large number of fighting types (would be enough for :Machamp:, :Hariyama:)
It's also an easy draft to "fix" as the second :lapras: or :mamoswine: get replaced by a fast and reliable/ good pokemon, its a strong team.

Worth noting that in some of the following battles I future-gen hallucinate :mamoswine: not having thick fat. This is a Hidden Ability but I'll note it as a definite error.

Team: :Zapdos:2/ :Mamoswine:1/ :Lapras:1

Battle 1: :Heatran:1/ :Victreebel:2/ :miltank:3
Good demonstration of why :Heatran: is a surprising (maybe it doesnt surprise some people) exceptional pokemon in the factory.
:Mamoswine: only comes in safely after something faints and given the type bias (fire/steel seen) and the fact that :zapdos: is just better than :lapras:, it's :lapras: who must come in and eat a potential stone edge.
:Heatran:1 does not directly threaten :lapras: although a resist switch does occur here. note that obviously :heatran: is well outside of an ice move range but it can be appreciated why "click another move" is the generally optimal way to avoid resist switches working.
Letting :lapras: drop is completely fine because I know that :heatran:1 wont be able to resist coming in vs :mamoswine:. this sets up a :mamoswine:/:zapdos: 2v1 which is exceptional odds to win.
:Miltank:3 doesnt win despite getting the fire punch crit on :Mamoswine: (although crit burn would have been an issue).

At the time my commentary mentions the light screen option (which would have worked out way better had I known the :heatran: set) but I think physical :heatran: punishes this and physical (set2) is the QC set that is a problem for mamo anyway.

:Heatran:1 is top tier as not only does the typing work so well but scary face (like twave) is an exceptional support move.
:Mamoswine: leaves mostly because a double water weakness is unacceptable. :Lapras: is worse but its something I'd much rather hold onto with a glaring weakness on the part of :zapdos:/:heatran: not being covered.

Team: :Zapdos:2/ :Heatran:1/ :Lapras:1

Battle 2: (24iv) :Charizard:4 / :Hippowdon:4 / :Kangaskhan:4
Because of the iv disadvantage, this isnt a speed tie in the opening and :charizard: will obviously drop.
brightpowder would be a problem but its not like i improve my odds by going :heatran: (:Charizard: will never click blast burn here so ancient power is incoming) as :heatran: also has to hit Ancientpower through brightpowder.
:Hippowdon:4 is an uncomfortable thing to run into after just swapping out :mamoswine:.
I think the switch here is obvious although the commentary at the time only briefly touches on it (I think perhaps a lot of players would want to keep :zapdos: in for a safe pivot into :lapras:)
With fire/flying/ground seen, :Zapdos: is hugely likely to contribute on the final mon. :Heatran:'s relative utility actually goes down slightly as the biggest matchups it checks arent possible.
The overlap on what I'd want :heatran: most for (fast grass; steels; normals) is largely covered by the :lapras:-:zapdos: duo and :zapdos: importantly prevents a pokemon like :lanturn: from clowning on me (this would still be a bad matchup to be clear but its one i can likely put on a 50/50 at worst)
Given I'm quite confident in the flash cannon + aqua tail roll, there's no way the much higher risk/ higher return heat wave is viable.
As mentioned in commentary, QC would be an issue, although its more of an issue to happen on :lapras: since if :heatran: gets QC i can always avalanche-ice shard
:Kangaskhan:4 isnt a problem

There is a small argument for Hippow4 > :Heatran:1 (I personally side on :heatran:) but for a poison spec the more kneejerky hippow >< :lapras: is an option too (as a "never lose now" option).
Long term the best team is to not swap at all and after giving both ideas a reasonable amount of thought i stick.

Team: :Zapdos:2/ :Heatran:1/ :Lapras:1

Battle 3: :Jynx:4 / :Roserade:1 / :Drapion:1
Seeing :Jynx: first is a lot of information on a poison spec as it largely means that :zapdos:-:heatran: handles almost anything (its basically just poison types with EQ that :zapdos: is needed for).
regardless the obvious :lapras: (it will OHKO :jynx: and takes almost nothing from ice moves) is the clearly correct play.
:Roserade:1 gets an unlikely stun spore which matters more than it may seem as :drapion: 1 reveals acupressure.
Full PRZ/ Acupressure atk is actually an ok outcome as it cannot boost past +2 (it will always see the aqua tail OHKO) and +2 is not enough for it to directly beat :lapras: or :zapdos:.
With hindsight I slightly prefer :Zapdos: first as the higher-variance attack (thunderbolt) is more important to use first in case a decision needs to be made later regarding ice shard.
Still the ordering does not matter much as :drapion: cant ohko either and is equally likely to acupressure or attack.
After Avalanche - no attack - curse is actually the safest option as the only way to lose now is via an acupressure speed and attack boost
The AI actually gets exactly this for a +4 atk +2 spe AND +2 def :drapion: which incredibly has won all 50/50 score 0 move splits and perfectly "outplayed" avalanche spam, although the fact that i did weave in a curse for safety is what saves this game as this prevents +4 drap from getting the OHKO and also means with avalanche spam I am guaranteed the KO. Only an additional acupressure into atk/eva would save this.
Without a curse (and the curse cannot be on the first turn) :lapras: can straight up lose. and with a late curse ice shard wont OHKO.

I highlight this as an incredibly close game, almost a 3v1 loss, and the kind of game where I can imagine another player losing very badly without thinking about roll permutations.

The swap here is a bait, I think :Jynx:4 is just too physically weak to take. While :drapion:1 is good it is worse than what i have.

Team: :Zapdos:2/ :Heatran:1/ :Lapras:1

Battle 4: (24iv) :Slaking:4 / :Donphan:4 / :Honchkrow:4
Truant gets special AI to boost giga impact's chance of being chosen. This - alongside how much hammer arm does - prevents a setup from :lapras:.
Scary face is almost correct but not quite. Flash cannon doesnt do enough, and if a move is going to hit then its a lot more important that i only rely on it once. Scary face would require scary face AND heat wave to hit (81%). Just heat wave only requires a single 90% move.
After heat wave, :zapdos: can light screen freely into tbolt. :Heatran: down sucks but its still a 2v2 with light screen up.
:Slaking: ignores this though by rolling a miss. This good luck is immediately offset by a QC activation on :donphan: though.
Stone edge is heavily incentivised here but Fissure is also against :lapras:. Stone edge is a lot better as it cant crit and avalanche ice shard should always KO :donphan:
:Donphan: activates QC again and picks Fissure but it misses. an hit was a very likely loss.
This is still somewhat "bad" though as without the doubled power, ice shard isnt a KO. Aqua tail is but its the 10% miss move i've tried to avoid.
ice shard outspeeds QC but doesnt solve the problem of fissure hitting so I am forced to aqua tail. With perfect hindsight ice shard better as :zapdos:2 always beats :honchkrow: and simply must dodge QC stone edge crit. but without hindsight there's no way intentionally playing into such a likely loss is correct.

:Slaking:4 is really good and has enough role overlap with :heatran: to drop it, although :slaking: in the back is significantly worse it also simply has a more even matchup spread which this team can exploit.

Team: :Zapdos:2/ :Slaking:4/ :Lapras:1

Battle 5: :Blastoise:2 / :Rapidash:3 / :Tentacruel: ?
Incredibly important not to get baited by matchup leads like :blastoise:. A torrent water move can drop :Zapdos: here and :Zapdos: wont OHKO :blastoise: without a +spatt nature.
The punish space for a :lapras: switch is tiny given shell armor.
slightly better than :lapras: is light screen though, which kneecaps set 1 and 4 while inviting 3 to bring itself into KO range via life orb.
Set 2 is maybe the worst outocme, but I decide its worth letting :zapdos: potentially tank a single torrent waterfall instead of risking multiple non-torrent, in order to bring :lapras: in and (more) safely remove this.
this is maybe the closest range of the run for ice shard to KO/ not KO. Ice shard has quite significant upside though of being a 100% acc move that is most likely to block punch and also potentially saves me from waterfall damage if that is clicked.
:Rapidash: coming in does slightly suggest sunny beam but its often a trap to get too baited into this logic as proved in the match.

Electric spec slightly spooks me here. :Zapdos: and :lapras: both arent great into it and :Rapidash:3 is on the bubble of being an acceptable pokemon.
With :Slaking: in the lead slot i could stick but without it there I have doubts on the team's ability to handle 2 decent electric sets (admittedly, electric sets mostly are below average).
The pressure on dropping :Lapras: is much higher as the non-lightscreen mon which has a genuinely bad matchup as opposed to :zapdos: which can "almost" beat some better electric sets.
:Lapras: is also a worse pivot-in on a truant turn if I know the opponent is an electric type which will bias towards their stab move vs :slaking:.
The combination of these reasons is enough for :Rapidash: although this would be a really hard decision on battle 2 or 3 of a round.

Team: :Zapdos:2/ :Slaking:4/ :Rapidash:3

Battle 6: :Ampharos:4 / :Milotic:4/ :Manectric:1
Turn 1 is a pretty hard decision as was expected in perhaps ~55% of scenarios for an electric spec.
As mentioned in commentary the play I _want_ to make is to get :rapidash: in, flare blitz, and then have :zapdos: finish it with tbolt. :Ampharos: is slightly too bulky to do this to.
I think reviewing the footage i agree with the overall decision on :slaking: and the tiebreaker is mostly how :rapidash:3 is just good enough for the job of taking out some random other electric type, alongside the huge upside of the outcome if :slaking: can somehow survive the next 2 turns mostly intact.
Set 4 is an ok outcome although I would obviously rather it miss thunder and 99 is lower hp than i was expecting :slaking: to be on.
The good news on low HP :slaking: is it increases the chance milo attacks for a "safe" :zapdos: pivot (as opposed to e.g. hypnosis)
The switch in vs :milotic: is another massive decision with a lot hanging on it. It's way better to kill with flare blitz and avoid a free truant turn but equally if blitz doesnt KO its fairly likely to be over.
Perhaps someone will analyse the exact pixel count but for me its often intuition which does guide something like this. Blitz definitely does around 60hp and milo looks almost bang on 30% (60hp). It's a risk not worth taking, i can reset truant if i have to (at least :slaking: can obviously EQ here)
:Manectric: is actually an easy thing to make a decision on. The special sets make :slaking: useless, and :slaking: probably survives 2 thunder fang from the physical sets, meaning in both scenarios the chance staying in for truant is best is fairly high (noting that physical :manectric: is unlikely to beat :rapidash: anyway).
Note that even against set 2 i get "1.5" drills if it does rain on :Rapidash: (it will not rain dance on :slaking: as thunder will OHKO 100%)

last hint is steel and one of the few types where i wouldnt drop :rapidash: for the old electric/water/normal core via :milotic:.
:Milotic: is pretty good vs steel types too. But so is :Zapdos:. With EQ in the team i would describe swapping the fire type here as a somewhat blind misplay and one :Lucario: could happily punish.

Team: :Zapdos:2/ :Slaking:4/ :Rapidash:3

Battle 7: :Lucario:2 / :Forretress:2 / :Absol: ?
:Lucario:2 is an ugly lead but does not get the crit it needs on :zapdos:. At least :Rapidash: is also faster in the back.
:Forretress:4 hits :rapidash: dangerously hard with rock slide (easily can crit OHKO). sacking :zapdos: is the safest play by far.
:Absol: can never 1v2 :Rapidash:/ :Slaking:
(3:04:00 in)

draft: :Raikou:2/ :Zapdos:1/ :Hypno:1/ :Manectric:2/ :Houndoom:2/ :Exeggutor:4
hint: -

Contender for one of the worst drafts in the run because of how limiting it is.
There are problems with all 3 electric sets although :Zapdos:1 is comfortably the best although in the context of the factory its barely above average.
- :Manectric:2 is a worse version of :Zapdos:1 and its MUCH worse and also just bad in general.
- once you get over how dumb physical riakou sounds, :Raikou:2 is OK. It's average.
- :Hypno:1 is a contender for one of the worst sets in the factory excluding those without any STAB attacks.
- Physical :houndoom:2 is a joke although it probably scrapes by as like, barely below the true median
- :Exeggutor:4 is a dubious set and more often than not a trap due to terrible speed and way too many weaknesses. that said we may need it here.

The temptation is to take the best 3 sets (:Raikou:2, :Zapdos:2, :Houndoom:2) but this is simply too weak into something like Rhydon or even something dumb like :Tangrowth:.
:Exeggutor: doesnt work well here. It's not an ice switch in, well its not an anything switch in if we're honest. but it can come in after a faint and either leaf storm or psychic a thing.

:Zapdos: is definitely the lead mon as the best set and the one with thunder wave support for :exeggutor:.

Ditching at least 2/3 of these will be necessary as i'd give this team a <40% chance of going 7-0.

Team: :Zapdos:1/ :Raikou:2/ :Exeggutor:4

Battle 1: :Kingdra:2 / :Charizard:1 / :Espeon:1
As stupid as this sentence sounds, I think it's worth preserving Egg on :charizard:. No water type in last reduces the chance of it mattering over :raikou:.
:Zapdos: is forced to twave :charizard: in order to prevent a potential Petaya-blaze sweep (:raikou: probably blunders into that range) on the thunder miss. If :charizard: won the speed tie, :raikou: would have been forced to return -> thunder fang and accept whatever damage or debuff came.
:Raikou: first is slightly more optimal in case of something exceptionally dumb happening like leaf storm miss + Reflect into a psychic crit.

:Kingdra:2 is not a great set and often is a yawn bot and brine finisher.
:Charizard:1 can be very good but can also be super high variance. It's a much riskier option that could pay off.
:Espeon:1 is arguably not even an improvement over :Raikou:2.

Team: :Zapdos:1/ :Raikou:2/ :Exeggutor:4

Battle 2: :Claydol:1/ :Ambipom:1 / :Empoleon:2
stay in vs egg is a surprisingly close call. :Claydol:3 is the immediate problem (for every decision) as even yache :zapdos: is not safely evading a 2hko.
If it is :Claydol:3 and i go egg, i will have to pivot back into :zapdos: through :raikou: and return to the original 1v1. This isn't great but it isnt too bad either as :Exeggutor: can contribute on a paralyzed mon later (plus if :zapdos: loses this 1v1 its a disaster already).
I took a double check on the live comm take and it turns out that :ambipom: can actually survive -2 psychic with -2 leaf storm although the roll is massively favored. Something i should have checked.

:Empoleon:2 is yet another dubious / below average set.
The defensive synergy of :empoleon:-:exeggutor: in the back is surprisingly ok although i would never rely on :empoleon: to contribute on its own.
This doesn't truly fix the team as there is still a lot of pressure on :Zapdos:1 to be cleanup as lead and this is a classic bad drafting problem, but there is no way to fix it and its not like i'd ever want to lead egg or :empoleon: anyway.

Team: :Zapdos:1/ :Empoleon:2/ :Exeggutor:4

Battle 3: :Flareon:3 / :Wailord:2 / :Shuckle:4
Unfortunately my mic doesnt catch up quite a lot of the comm here because its a surprisingly instructive match.
Turn 3 switch vs no switch is a pretty big one here. And :Raikou: or :Empoleon: were both options as return "probably" KOs. :Empoleon: is safer but also runs into the issue of being out vs the next pokemon.
The :wailord: matchup is instructive for anyone who isn't aware of how powerful decomp knowledge is. By knowing exactly when :wailord: will/ wont rest, i can wait on leaf storm until i know its guaranteed to lose so long as it hits.
:Shuckle: cannot touch :empoleon: but it can significantly stall it.

These pokemon all suck

Team: :Zapdos:1/ :Empoleon:2/ :Exeggutor:4

Battle 4: :Toxicroak:1/ :Snorlax:4
Very rare case where :Snorlax:4 can be discovered for certain from item knowledge alone.
I do like the decision to get out of :snorlax:, preserving :Zapdos:/ :Exeggutor: over :Empoleon:. Potentially better is the decision to sack :exeggutor: directly, and then going :empoleon: and then :zapdos: but this fails if :snorlax: uses d-edge on the 2nd egg turn.

:Snorlax:4 is significantly better than :Empoleon: and I do not want to stack a fighting weakness and leave the team open to :lucario: and friends, so its :empoleon: leaving.

Team: :Zapdos:1/ :Snorlax:4/ :Exeggutor:4

Battle 5: :Manectric:3 / :Pinsir:3 / :Starmie:3
Leaf storm being 10% miss is just generally a panic. It doesn't happen here but its easy to appreciate how bad things can go if it does. :Exeggutor: doesnt have the bulk or speed to function with a set like this.

:Starmie:3 is a completely different tier to :Exeggutor:. The team now passes into "good" territory if :zapdos: isnt forced to single turn thunder.

Team: :Zapdos:1/ :Snorlax:4/ :Starmie:3

Battle 6: (24iv) :Politoed:4/ :Charizard:4 / :Toxicroak:4
Shoutout to the first 3 turns of this match achieving almost nothing. The only thing :Zapdos: really had to achieve was prz + a single thunder hit for :snorlax: to EQ it.

Under other circumstances :toxicroak:4 would enter the team but I do not want to put it is as a lead and i like the :snorlax:4 twave backup too much to quit on it.

Team: :Zapdos:1/ :Snorlax:4/ :Starmie:3

Battle 7: (24iv) :Wailord:4 / :Mr. Mime:4 / :Snorlax:4
:Zapdos: cannot safely rain dance without a potential blunder into QC. the 30% into 9% miss/ double miss is more attractive than that as an option, and if not then :Wailord: wont hyper beam :starmie: from full anyway.
:Mr. Mime: gets a funny crit here which becomes less funny when :pinsir: comes out and threatens to potentially guillotine :snorlax:.
This is however a huge mistake!
The final 2 turns are played awfully. Switching in :snorlax: and then switching out of CC always wins as :Starmie: will always KO a -1 spdef :pinsir:.
This is the kind of exceptionally stupid mistake that should not happen and is something I'm definitely blaming on 4 hours of playing straight along with distracting myself about how the crit on :zapdos: removed the simple win.
I did at least have the presence of mind to Crunch and not double edge or zen (both could definitely risk a loss / tie (tie is a loss))

In fact even after the match live commentary misses this and instead focuses on the :Mr. Mime: crit switch option.
I think if I catch this at the time then I stop instead of continuing. I should have stopped anyway.
(4:09:00 in)

mic really doesnt pick the commentary up well so i'll pick up the slack here.

Hint: -
Draft: :Lopunny:2/ :Gastrodon:3/ :Exploud:2/ :Venusaur:4/ :Dewgong:1/ :Typhlosion:3

- :Lopunny:2 is a never pick. Genuinely.
- :Gastrodon:3 is ok
- :Exploud:2 is bad
- :Venusaur:4 is ok/bad
- :Dewgong:1 is bad
- :Typhlosion:3 is ok/ slightly above average

This probably is the worst draft of the run although it doesnt look like it at first glance. The problem is i am basically forced to pick a somewhat trapped team.

:Typhlosion:3 is the best pokemon and the lead although ideally it never faints before getting out.
:Gastrodon:3 is bad but at least has no chance of having to 1v1 a grass move.
:Venusaur:4 is probably a forced closer with frenzy.

This team is a "trap" because fixing the lead without introducing another weakness is hard. :Venusaur: cannot be replaced before :Gastrodon:/ :typhlosion: unless a weird type spec comes up. :Gastrodon: pretty much must be replaced with either a dragon or water type.

Team: :Typhlosion:3/ :Gastrodon:3/ :Venusaur:4

Battle 1: (24iv) :Sceptile:4 / :Lanturn:4 / :Slaking:4
Pretty simple match. If Frenzy misses :Slaking: loses the final 1v2 because it will bait itself to EQ into Shuca which would allow :typhlosion: to EQ 3 times into a blaze flare blitz.

It says a lot that i would happily swap my entire team for the opponent's.
Unable to do that I settle with :Venusaur: >< :Slaking:, getting rid of the worst pokemon for the best one.
I would definitely prefer to lead :Slaking: but I have partially committed to :Typhlosion: at this point and :Slaking:/gastro/venu has some glaring speed issues.
The next hint being flying also pushes this even more.

Team: :Typhlosion:3/ :Gastrodon:3/ :Slaking:4

Battle 2: :Vespiquen: ? / :Mamoswine: ? / :salamence:2
The question of what to do on :mamoswine: is somewhat close however i do agree with the slightly more hindsight view of just attacking with :typhlosion:. Unlike some other decisions, I'm not going to label this as an objective mistake though.
Keeping :typhlosion: does help against a fairly long list of flying types - and it is true that if :Mamoswine: survives blitz the situation is uncomfortable with :slaking: probably being forced into truant vs a 3rd pokemon.
However :salamence:2 is a uniquely bad punish for this play as it makes the :typhlosion: keep irrelevant and the EQ damage on :slaking: matters massively.
As mentioned in commentary the main silver lining of this unlucky/bad outcome is that 3x DD is impossible, meaning that at worst Gastro gets to have a chance at surviving.
However there is no doubt that a -1 :slaking: facing a :Salamence: which just got a free DD and will already ohko :typhlosion: and survive stone edge is a dreadful outcome overall.
+2 dragon claw is a ~30% roll to kill gastro, factoring in crit. This is another close call. Unlike some other games that were close, I'm not sure I can call this a misplay though.

:salamence:2 works both in the lead and in the back and its an opportunity to get rid of :Gastrodon: and finally have a team made of above average sets.

Team: :Typhlosion:3/ :salamence:2/ :Slaking:4

Battle 3: :Blastoise:3 / :Dragonite:3 / :Cresselia:3
:Blastoise:3 is an ugly lead. :Typhlosion: does so little to :blastoise: in general that keeping it in is a super low value play.
Sacking :salamence: for damage and to scout is actually ok here. :salamence:2 is not the carry that you may think it is, its also not really a setup sweeper. its most common play pattern is something like this.
But against :blastoise:3 since continuous intimidates and LO recoil build up, pivoting back through :typhlosion: and :salamence: works well. This is worthwhile because :Typhlosion:'s HP doesnt really matter that much and :salamence: was going to die with the other play regardless.
I am caught out by the "Revenge" (avalanche) ai here though, which is a blanket +2/-2 with a 30/70% chance respectively. (It uses avalanche on :typhlosion:).
This is doubly bad because a low HP :salamence: means :blastoise: will behave more randomly.
Cutting the losses here (thunder fang and it will faint to life orb) is the best option left, a bit of good luck also keeps :Salamence: around for one more matchup.
A naked :dragonite: is way too scary to switch out of. :salamence: always outspeeds and dclaw is always good chip for :typhlosion:. As mentioned in commentary, low HP works in my favor here as it makes Dragon Dance impossible to select.
:Cresselia:3 is the better outcome as it lacks healing which could have been an issue for :Slaking:.

Although :dragonite: is tempting and double dragon often works well, :Cresselia:3 is a ridiculously strong pickup and compliments the team well. It's not usually a lead but that's fine.
Ground type is yet another reason for this swap.

Team: :Cresselia:3/ :salamence:2/ :Slaking:4

Battle 4: :Hippowdon:4 / :Umbreon:3 / :Mamoswine:2
Not much to comment on here except to say that fire fang would have been a needless risk (sitrus :mamoswine: would have an outside chance against damaged cress + :slaking: as Giga impact/ hammer arm as I would probably have to Earthquake into energy ball)

The commentary here is cut out by the mic but I am essentially saying a team like this is about as good as it gets with bulkier pivots alongside 3 strong mons. The replacement option would be :Salamence: but it would take an exceptional swap option to drop :Salamence: here.

Team: :Cresselia:3/ :salamence:2/ :Slaking:4

Battle 5: (24iv) :Ampharos:4 / :Scizor:4 /
Commentary cuts out but I basically say that trading Cress for :Ampharos: is fine with me and much preferable to risking one of the other 2 taking a ~21% thunder paralysis (or worse, crit).
:ampharos: is unfavored to win even ignoring the spdef drop chance, as it must hit multiple 84% thunders and it must also either crit or prz with at least one.
:salamence: covers :scizor: very well just like it covers most physical attackers well. Even the worst case scenario of :Salamence: getting crit or missing, We'd still be getting :slaking: in on a -1 :scizor:.
I must admit off the top of my head i have no idea why :magmortar: would focus blast a 100+hp :salamence:. the switch was definitely intended to come in on either overheat or psychic but mostly just to reset confuse so :Salamence: always cleans up.

None of these pokemon are swap candidates. Despite the synergy I would consider :Slaking: >< :Scizor: a huge downgrade in flexibility.

Team: :Cresselia:3/ :salamence:2/ :Slaking:4

Battle 6: :Probopass:4/ :Shiftry:4 / :Kingdra: ?
Even against -1 attack :shiftry:, I do not want to DD more than once as its very likely to be an unnecessary risk. a crit aerial ace into a sucker punch next turn would be embarrassing and mean a truant turn :slaking: is forced to face the final pokemon.

With a fire type specialist next plus it being the last match next, there's actually a reason to take :kingdra: even though the set is unknown.
Commentary mentions (although with the cut outs its hard to hear) that actually the most threatening list of fires dont run through :salamence:/ :slaking: anyway, and if it is :kingdra:2 then that's a significant downturn in generic use vs the 3rd pokemon.
While true, I could see taking :Kingdra: regardless as even :Kingdra:2 is likely to contribute well. :Kingdra: would go in over :Slaking: mostly out of respect for how a few special fire sets could ohko :slaking: with a crit 120+ stab move. This leaves few possible losses open but the most prominent is probably :Garchomp: or Lati@s after :Cresselia: takes any damage; so in this case :Cresselia: would have to leave the field on t1 if it was a fire type.

Team: :Cresselia:3/ :salamence:2/ :Slaking:4

Battle 7: (24iv) :Starmie:4 / :Rapidash:4 / :Ninetales:4
Basically free, particularly after :cresselia: was perfectly in megahorn bait range (only :Rapidash:4 had any real chance of doing something here).
At the 4 hour mark i criticised the decision to continue but 5 hours in I think it was just wrong.

(4:55:00 in)

draft: :Jolteon:3/ :Articuno:1/ :Poliwrath:4/ :Electivire:4/ :Granbull:2/ :Honchkrow:4
hint: normal

- :jolteon:3 ok
- :articuno:1 ok
- :poliwrath:4 slightly above ok but also variable
- :electivire:4 amazing
- :granbull:2 terrible but it does have intimidate which puts a floor on how terrible it can be.
- :Honchkrow:4 is slightly above ok but needs to call suckers quite often to work that well

:Electivire:4 is one of the best leads as it really does make the enemy behave so predictably while also winning a lot of 1v1s.
However the lack of a clear team actually means it is benched. Instead it is :Jolteon: who works better as a temporary lead via yawn more than anything else.
This also makes it a lot easier to keep :Electivire:4 around on the team during swaps.
:Poliwrath:4 is not a set I like but it is objectively the next best thing. A :honchkrow:4 fan could make a case for it but personally I do not like sucker punch and I do not trust it to not fall over.
The ground weakness is ok, particularly knowing this is a normal trainer.

Team: :Jolteon:3/ :Poliwrath:4/ :Electivire:4

Battle 1: Mismagius3 / :Tauros:4 (anger point) / :Blissey: ?
I think despite mismagius2 having lum that yawn would have been the better decision on turn 1, since :electivire:4 outspeeds mismagius2.
The low hp makes DBond super likely and it forces a turn 2 yawn. Although straight up trading :jolteon: for mismagius is ok, yawn would effectively achieve the same thing.
What I do agree with is the decision to bring :Jolteon: back in on the first turn of sleep in able to repeat the better lead. Resisting the urge to just waterfall is optimal over a larger set of scenarios and the switch risks nothing as its a yawn-induced sleep.
:Tauros: in next is terrifying as if its set 4 and uses return, i lose unless i yawn. but if i yawn and its set 3 then i also very likely lose (or effectively lose i.e. low HP final mon in a 1v1) to outrage spam.
It is set 4 but it does not get click return. Stone edge is actually more likely due to the flat increase score chance to all crit moves.

:Tauros:4 is the obvious take. Mismagius3 is a good mon but :tauros:4 is borderline broken even without intimidate
:Poliwrath: leaves first for sure. While its a better set than :Jolteon:, :jolteon: works very well in able to safely bring :tauros:4 in after a yawn

Team: :Jolteon:3/ :Tauros:4/ :Electivire:4

Battle 2: :Honchkrow:2 / :Slaking:3/ :salamence:1
Between super luck, sucker punch and even torment, it's not really an option to just use thunder and even rain dance into thunder is dubious. Even yawn is blocked as it can be insomnia.
Instead the far safer play is certainly to bring :Electivire: in against :Honchkrow:.
:Honchkrow:2 is not a threat though as it will not superpower at full health and both :jolteon: and :electivire: will obviously OHKO it
:Slaking:3 is a really bad follow up because as mentioned immediately in commentary, its jolly (:tauros: is adamant).
The safest continuation to avoid losing to a 1 turn sleep is to Thunder and then yawn (if thunder misses). In this scenario :tauros: never loses as the first sleep turn is guaranteed to overlap on non-truant.
This is still ok as the last pokemon is flying type and locking into return is ok there. The punishes (:drifblim:; :Aerodactyl:) are a lot rarer than the things destroyed by return.
Note that this is even safer than rain dance into thunder as it has the bonus of if 2/2 thunders hit then i get a 2v1.
:Jolteon: does not miss a thunder and :Salamence: comes in. Yawn is a lot safer here despite the chance of lum, as set 2 is the biggest threat. It is instead set 1 which is easily handled if asleep.

:Slaking:3 is better than :salamence:1 in the lead slot (the slot they'd both be going into)

Team: :Slaking:3/ :Tauros:4/ :Electivire:4

Battle 3: (24iv) :Garchomp:4 / :Floatzel:4 / :Luxray:4
:Garchomp:4 is not that bad as :tauros: is faster but its certainly a small punish for not taking :salamence:1.
:Slaking:3/ :Tauros:4 are somewhat reliable for the closing matchups although I'm definitely put on the back foot, as is expected against :Garchomp:
Sacking the :Slaking: is also an ok option and perhaps I overvalue the +20 speed it has over :electivire: as well as the fact it does almost double damage with its stab. then again maybe not.
:Floatzel:/:Luxray: arent the backline the AI needed to win this though.

:Garchomp:4 is basically always a must take even with the slightly painful IV penalty.
Because the normal types are so fast I'm happy to leave them both in the back. :Garchomp: joins over :electivire:.
Putting :Garchomp: in the lead slot is also fine. :Slaking: encourage a lot more switching which is closer to my playstyle, perhaps other players would have gone for chomp in the lead.

Team: :Slaking:3/ :Tauros:4 / :Garchomp:4

Battle 4: :Staraptor:2 / :Snorlax:1 / :Flareon: ?
it takes me a few seconds to realise it cant be band because of :tauros:. that leaves set 1/2 only which both suck.
Still :slaking: doesnt want to stay in at -1 and :tauros: handles the cleanup. Set 1 is hugely likely to double team here which is a :tauros:-prefered matchup for obvious reasons.
Ironically the safest 2 options possible in the draft (:jolteon:/:electivire:) just left.
:snorlax:/ :flareon: are doing nothing into a team with this much raw damage output.

:Tauros: swap for :snorlax: is a small consideration. if thick fat was known then i'd perhaps do it.

Team: :Slaking:3/ :Tauros:4 / :Garchomp:4

Battle 5: :Jynx: ? / :Blastoise:4 / :Venusaur:4
:Jynx: remains a joke with its slow sets that faint to almost anything. It is just "bulky" enough that :slaking: cant aerial ace to dodge lax incense but i'll note that this is still close.
Turn 2 is mindless, there is no other word for it really. While its true that 3/4 :Blastoise: are not a problem, set 4 would obviously focus blast and its not a roll on :tauros:.
Staying in with :Slaking: was correct into every set, to follow up with a return. This is a straight up mistake.
Instead of facing a pretty simple endgame 2v1, I'm instead in a very difficult spot. :Garchomp: cannot come in as it will drop to ice beam and 80% truant'ed :slaking: doing the final 1v1 would be stupid.
So :Slaking: must come back in but it obviously will not OHKO and if I use return then I'm 100% giving it torrent hydro cannon (destroying :slaking: in the lucky event of a focus blast miss).
But no other attack sets up a guaranteed :Garchomp: kill. So I'm just blindly falling back on a :Garchomp:4 blind 1v1.
:Slaking: instead crits :Blastoise:. I'll note that :Garchomp: should safely beat :Venusaur:4 in a 1v1 since the ai will never frenzy plant or hyper beam or synthesis on turn 1, leaving the door open to a simple outrage 2hko. This doesnt make the misplay alright though.

no swap, no viable option anyway.

Team: :Slaking:3/ :Tauros:4 / :Garchomp:4

Battle 6: :Dewgong:1 / :Golem:2 / :Skuntank: ?
despite playing much faster here it's played a lot better. Even spending multiple minutes now thinking about the :golem: decision map i agree with the ultimate decision.

no swap makes sense again

Team: :Slaking:3/ :Tauros:4 / :Garchomp:4

Battle 7: (24iv) :Hypno:4/ :Muk:4 / :Lapras:4
Another misplayed game, :Lapras: gets lucky but its definitely correct to bring :Tauros: in first before :slaking: as all accurate combinations here will be 2hkos.
The play in the game turns out to be correct but for the wrong reasons as the return roll is clearly high enough for AA to KO later, so bringing :tauros: in on truant guarantees the win.

More than any other, this round was carried by a strong team as the decision making was off. Not massively off, but off in a way that definitely matters.
This round still makes me cringe to watch back. I suspect if viewed without my commentary (either live or here) perhaps the decision making looks fine to a casual observer but its deifnitely not.
Suboptimal plays are fine but this goes beyond that.

(5:41:00 in)

draft: :Heracross:2 (guts)/ :Lapras:4 (shell armor)/ :Latios:1/ :Toxicroak:4/ :Garchomp:3/ :Shiftry:3
hint: -

to clear the air on this draft because at the time my brain somewhat short-circuited.

:Heracross:2: good
:Latios:1: good but no better than that
:Lapras:4: amazing, particularly with shell armor
:Toxicroak:4: great
:Garchomp:3: great/ arguably better
:Shiftry:3: kinda bad but not like the grass type was ever it.

The take here should actually be: :Garchomp:3/ :Lapras:4/ :Heracross:2 OR :Garchomp:3/ :Lapras:4/ :Toxicroak:4
I would say every other team is actually wrong. The aversion to "risky" :garchomp:3 set is a human error when it is much more likely to close and work when needed compared to the team taken
While it looks safe, this is wrong:
:Latios:1/ :Lapras:4/ :Heracross:2
As this fundamentally fails in the check for leads and exit points. If I wanted this team because I never want to touch :Garchomp:3, then :Heracross: should be the lead.

I think with less fatigue i never make this mistake. Regardless, its the decision I make.

Team: :Latios:1/ :Lapras:4/ :Heracross:2

Battle 1: :Manectric:2 / :Muk:2/ :Registeel:4
It's important to underline that :Garchomp:3 being the best lead isnt because i can see that :manectric:2 was coming up, but yes this would have improved the match significantly.
Note a further mistake; under rain, Surf does more to :muk: than psychic. While visually it does look like modest :lapras: should make the KO with both there's no reason to click psychic here.

While :Registeel:4 is worse than :Heracross:2, I think for defensive synergy purposes it should probably be taken.
at the time I'm distracted by the Steel type specialist but this is battle 1 and the focus should be more on longer-term winning propsects.

Team: :Latios:1/ :Lapras:4/ :Heracross:2

Battle 2: :Scizor:4/ :Aggron: ? / :Typhlosion:3
It's probably too hindsighty to say that :Lapras: switch on turn 1 was better.
Overall the match was played correctly on key decisions.
I will once again underline that this match is a demonstration of why lead :heracross: is better than lead :latios: for this team, despite the twave assist. This really is not hindsight, its just not surprising that examples of the suboptimal lead choice immediately start popping up!

Pause here and really think about what you would do in this scenario.
next opponent: no type spec
:Latios:1/ :Lapras:4/ :Heracross:2 with the option of :Scizor:/:Aggron:/:Typhlosion:
The answer is incredibly obvious to me and I think 100% a product of fatigue. :Scizor: is better than :Heracross: and the Dragon/Water/Steel core works. The swap should definitely be :heracross: for :scizor:, or alternatively just no swap. No swap has some perks too, particularly regarding immediate damage and small speed optimisations.

Instead I swap :Lapras: out for :Scizor:. I know why i did this; because :scizor: + :latios: is a nice duo, but despite how strong this team is after twave support, and despite the fact that :heracross: frequently does get a bulkup setup in games, :lapras: fills a pretty important role. It's the worst drafting decision of the run.

Team: :Latios:1/ :Scizor:4/ :Heracross:2

Battle 3: (24iv) :Froslass:4 / :Espeon:4 / :nidoking:4
Although I don't comment on it in live coms, I do immediately realise how badly :Lapras: is missed when i see :froslass:4 come out. This is a 1v1 which is completely trivial for :lapras:.
Instead :Scizor: takes a hefty ~50 damage from blizzard and should take more but it misses.
:Espeon:4 is never setup fodder and must be removed. If forced, :Latios: could CM once into dragon pulse but I am sure without checking that this would not be a guaranteed OHKO.
:Scizor: takes another large chunk from psychic and ends up at 29hp, although one could argue that 49% of the time it should be dead from 2/2 blizzard plus psychic.
Last in is :Nidoking: and :Nidoking:4 somewhat embodies the exact problem with this team.
:Scizor: does beat :nidoking:. with more starting health. A classic problem has come up here where :Scizor:'s duties were simply overloaded as it must perform as a check to too many pokemon. This is alongside other issues:
1. :Latios: and :Heracross: are actually to slow to function as switch in revengers
2. The team is just too slow in general
3. :Heracross: should always have been the lead. :Heracross: vs :Froslass: is actually a 1v1 that :heracross: stays in and likely wins. After that the match is a straightforward win.
4. :Lapras: provides huge support that is perhaps comically underlined by this exact opponent.

Regardless, its time to play the :nidoking: 3v1 out.
I suspect the play looks fine to most people (perhaps you can spot the suboptimal decision - its something i notice in live comm also but i notice it too late).
- Rivalry :Nidoking:4 (:scizor: is female) will only ever EQ here, but also will always outspeed and OHKO :scizor:.
- Poison Point :Nidoking:4 can both Megahorn and outrage here. These moves would be very unlikely (only 1/4 to see the KO at 29hp).
- With any ability, :Nidoking:4 is incredibly likely to OHKO :latios:1. :Nidoking:4 is also faster than :latios:1.
- :Heracross:2 (female) vs :Nidoking: is more complicated, but the summary is that because :heracross: is slower, :nidoking: will be favored.

My first intuition on what to do is the following:
1. :Scizor: isnt contributing anything and any other pokemon taking damage to save it when its very likely to faint to EQ anyway would be dumb.
2. Therefore :scizor: is sacked this turn. We click Iron head as this would win the game if :nidoking: decides to not-EQ and misses the KO roll on :scizor:.
3. :Latios: comes in next, since although its unlikely to do anything, a megahorn miss (megahorn is 50% to be chosen) is an instant win with psychic.
4. :Latios: coming in also provides the benefit that if :nidoking: uses outrage, it will remain locked in outrage to :heracross: and then self-confuse. This provides a very realistic winning scenario.
5. The decision to bulk up or not on :heracross: can be made at the time and isn't affected by earlier baits. It's obvious :heracross: has a chance, particularly with guts and bulk up and since its oppositely gendered.

Sounds good right? Wrong, there is a far more likely way to make sure that :Scizor:/:Latios: contribute:
1. :Latios: should be switched in immediately on the (likely) EQ. If switched in on outrage this is even better as it means :Scizor: can come in and iron head.
2. If :latios: survives then :scizor: should come in again.
3. If outrage is chosen, :scizor: is sacked. a 2 turn outrage means :latios: comes back in with a huge ~57% chance to win immediately.
4. If megahorn is chosen, :Scizor: is now sacked as this is yet another 1/3 chance to fish for an early outrage. If no outrage, :latios: comes in. if outrage, :heracross: comes in.
5. If :Scizor: takes no damage (megahorn miss), then repeat step 1.

The second path is better but its very narrowly better. Anyway, the first decision tree is punished as :Nidoking: does not outrage on :latios:.
It's then a straight up 1v1 with :nidoking: vs :Heracross:.

There are 2 main options here:
1. :Heracross: attempts double bulk up.
- If the first turn reveals anger point :Nidoking: and :nidoking: chooses poison jab, then :Heracross: uses CC on turn 2. This wins if CC crits or if :nidoking: poisons :Heracross: with poison jab (on the second turn), because a +1 guts CC will OHKO :nidoking:.
- if the first turn reveals anger point :nidoking:, but :nidoking: uses outrage, then :heracross: will bulk up again into CC, as a 2 turn outrage into a self hit will always win as :nidoking: does ~40hp to itself.
- If the first turn reveals rivalry :Nidoking:, then :Heracross: can survive 3x 120 base power moves and double bulk up can win. Double bulk up must be commited to and if :nidoking: uses poison jab twice but only poisons the second turn, unlucky.
- If the first turn reveals rivalry but :nidoking: immediately poisons with poison jab, then :heracross: wins the next turn with +1 guts CC.
- General note that rivalry + turn 1 outrage with bulk up is a hugely likely win as this will almost guarantee :nidoking: self-confuses (2 or 3 turn both can work) if no crit against :heracross:.
2. Night slash into CC.
- This never wins against poison point unless :nidoking: poisons :Heracross: and then night slash crits.
- Against rivalry, this has a similar wincon to above if :nidoking: chooses outrage on turn 1, but has the additional benefit of :heracross: critting night slash or CC.
- Against Rivalry + poison jab on turn 1, this can also win if poison jab #1 (note; would be a roll for :heracross: to survive) or poison jab #2 poisons :heracross:. :Heracross: only needs a single guts boosted CC after a regular night slash.

The first option is _slightly_ better, although either one of these is acceptable. The overall win chance is surprisingly decent.

I go for Bulk Up.
What happens is that :Nidoking: reveals its rivalry, through a crit + poison poison jab. To add insult to injury this was a high roll crit so :heracross: just immediately faints (although note that after the crit its completely over).
the win chance for a turn 1 crit was almost 0 anyway (night slash crit after poison could have won but I'm pretty sure the bulk up line was optimal).

Losing didn't bother me but the mistakes did. :Garchomp:3 and :Lapras:4 can both save that fight. :Heracross: lead would have prevented the loss too. The misplay on :scizor:/:latios: before :heracross: also could have mattered.
Most of all I dont see myself making these mistakes (let alone all 4) if fresher. There was no reason to continue the run and no reason to feel so safe with what was objectively a pretty subpar team (:latios:1 and :heracross:2 are good but nothing special).
Ultimately I'm not fully happy. I didn't sink so many hours into the factory over a decade after I first beat it to get the record, I wanted to prove that more is possible and failing to play at the required level is the reason why the factory has not had strong records in the past (to be clear I'm including my own struggles to get gold in that group).

I'd much rather have got 85 wins but done a writeup where I could not criticise a single match.

Perhaps the play sessions were too long but also perhaps without committing to longer play sessions I never actually get a solid run going before losing interest again. At some point things like this become more about handling real-life sanity than optimising just around a game.

Still i guess i've proved my point, and this isn't even as slow as a player could go. Honestly every decision should probably take 2-3x as long and since my reliability on mental calcs isnt 100% i may as well actually do all of them properly. I don't know if that's fun anymore though, at least for me. I would really like to beat this run. Just simply so that the record can be of stronger play. I'm not sure I can do this, and not in a reasonable amount of time. Even when I'm going full sweat mode I still have around a 1.5-2% loss chance based on the hours of streaming and offline footage I have.

What does seem correct is for factory to finally be somewhat broken. It took the decomp for it to happen but at least the record is well out of Thorton's shadow.

If you want to beat this, you probably can. Read my guide and then improve on it/ play better than i did. Just know that it's a long road. 300 hours seems like a lot, I would estimate I'm well into the thousands on gen 4 and I probably plateaued/ stopped properly improving long ago. A lot more is possible.


I've also broken the record in Factory Open Level singles (100) and Open Level doubles (79). Both of those are on my Twitch/Youtube and I will write those up at some point. I am much more pleased with the Open Level singles record despite it being "a lot" shorter because I feel like my mistake frequency in play was a lot lower. Perhaps I'll change my mind when I review the footage. Or perhaps factory records are always kind-of destined to be luckier runs due to how important draft rng is.

It's unlikely I ever actively push for a record in Factory again. I can see myself playing it again on livestreams and doing showcases, I'm not sure I have the internal motivation to keep going for hundreds more hours.
 
Last edited:
Factory guide/ bleh/ Tierlist

Brief contents of what this guide goes over:
- preamble
- the brain capacity issue
- factory pinciples and overview
- tierlist of sets
- general typechart analysis
- leads
- Team composition in drafting
- how to evaluate sets
- common pitfalls
- coverage rant
- playing against the in-game AI guide
- notes on 28+ streaks and Thorton
- closing thoughts

The gen 4 factory has long had a reptuation of being too random/ too hard/ cheating against the player.
As someone who played extensively in the early 2010s I also strongly held this belief if simply due to how long my old PB of 56 took me.

However its now 2024 and a completely different era of knowledge. I think a lot of the old excuses have to be laid to rest.
A quick list of things that exist now but didn't before (at least from my perspective)
1. The full decomp of the Gen 4 AI procedures, move and switch logic. This is __the__ major difference maker.
2. A proper rip of the factory. Maybe this also existed, I wasn't aware of it back then. Knowing opponents IV tiers, sets, elevation logic, etc is huge - particularly pre-49 but also post-49.
3. The general availability of easy-to-use damage calculator websites as well as showdown itself and adjacent tools.
4. A better understanding of the game in general.

I also think a lot of previous runs/ write ups/ recorded footage fall short - or at least I presume they fall short, a lot of the write ups dont detail what important decisions were actually made. I want to bold a lot of these and ask either "how did you win this battle" or sometimes "how did you not win this battle since X"

What is actually possible - both in terms of drafting and play - is way higher though. More than the record being low this is what "bothers" me, because the factory is perhaps the only facility where the player is completely alone with their ability to play. Is luck involved? Certainly. But luck doesnt make the factory less skillful, it makes it more. You are forced to choose between certain outcomes and complex probabilty webs a lot.

Still, you should be prepared to lose in the dumbest ways possible. As an example; this was my first loss after returning to the factory for the first time in years: https://streamable.com/0dwjru


The Japanese record is 90 as far as I can tell:
This isn't about specifically calling this record out, but the play is massively different to what I would do - to the point where I feel like we are both idiots. I at least rarely understand or agree with drafting choices, and frequently suspect that no precise damage calcs are going on (although, i'll also cricitise myself for this at points. Don't worry, this isn't an arrogant tirade but more of a "wow humans are really bad at this arent they").

In terms of English Factory footage: I'm not really aware of any at a high level. Werster has recently been speedrunning it but this isn't the place to go for ultra-long streak advice (and I'm sure he would agree that he would play differently if he was not trying to speedrun). I wish there was more.

I guess there's also my stream and some of my commentary as well as my streak's write up now. I don't know if this still views as alien to other players as the Jpn record above does to me. Still, the write up for my 156 winstreak is supplementary to this and crammed with examples of what I refer to here.
I think the biggest problem with the factory is just that a lot of optimised decision making is a bit too algorithmic.
We like to have simple heuristics and guides and go off experience but the Factory is kind of (unintentionally?) designed to leech off this.
There are some shallow surface-level truths in the factory (fast pokemon are good, steel is a good type, flying and water are common types). Anyone not familiar with these basic first steps aren't going to get far.
At the same time, adhering too closely to a set of defined rules is also bad and probably even worse. At best such a player would be stuck bashing their head into a wall, doing things that mostly work but fail at critical points. Without challenging any thought processes it may be almost impossible to ever improve.

The factory lulls people in by pretending to be similar to PvP or romhacks or nuzlockes or whatever. It's different though and you don't want to break your intuition for how the game works but you also don't want to ever believe you're beyond learning a new lesson.
Essentially, a bit like the details in more complex statistical models and scenarios, your own brain can be an enemy here. What does "the enemy team is largely randomised sets pulled from higher BST tier4+ pokemon" really mean to you?
I can point out facts, like how compared to 6v6 PvP gamemodes, certain types are way more or less common, and how important this is. Or how without or with the introduction of common/ uncommon moves from "pokemon in other contexts", some sets get a lot better or worse.
I can also say that random flies in the face of logic to how you might feel their team should be structured. Surely they have a ground type after seeing a flying type? Such a motif is common in PvP. Or ingame, (particularly with type hints), if we're told they're a bug specialist, surely this means they can't send out Moltres first? You might kick back at this and go "I know that, I know what random means". But do you? When you considered any lead could show up, did your mind flash through all 600 sets? Surely not.
My heuristics and mental models based on this are flawed, and so are yours. Particularly if playing "fast" (for context, taking less than 5 minutes per turn is fast).
There is an insanity and contradiction baked into this guide. You will never finish learning about how to play the factory optimally.

While I'm on the "check your ego" topic I want to underline 2 very common mistakes that stronger players make too often:
1. Complexity bias / Pro-innovation bias is dangerous. Caught between two options - one being to click your strongest move, and the other being a complex series of pivot switching, taking advanced of a traced ability immunity, to attempt to PP stall a move - which appeals more? Inevitably it seems that once the complex play is spotted that its too alluring to turn down. Surely the play that a noob wouldn't see is better? But no. Beware random. There is no karmic RNG jusice awaiting a "harder to see play". The problem in front of you may have no solution that leads to a victory. Just be objective, whatever that means.
2. These aren't the droids you're looking for. Do not assume to know how a factory set operates just because you've used the pokemon in other contexts. You're probably familiar with Skarmory sets that run heavy HP investment, beware just how different some of these 0HP skarm sets are when it comes to tanking.

I often wonder what my own badly approximated heuristics have meant for a lot of historic dumb losses. My feeling is that they're perhaps to blame for maybe 30% of them, maybe more.
This should be intentionally straightforward at the start but its what we'll build a lot of the guide on going forward.

Lets go over what a typical factory win for the player looks like:
1. The player plays out a series of 1v1s, where the first 1v1 is won and then the switch in is chosen tactically. This continues to what should be a "1.5v1" at the end. The player's advantage does not necessarily come from type matchups but just from the fact that the player usually drafts better pokemon as the AI's team will be random.
2. The player's lead builds a substantial edge that is easily converted. Perhaps it doesnt outright KO 2 pokemon, but it may get a KO and then heavily cripple or damage the next. From here its a 2v1 scenario where the player should have a large advantage if their team is somewhat logical.
3. The player is put at a disadvantage early (either losing the lead matchup or being forced to switch out of it) but is able to recover either in thanks to good play, a later matchup advantage, or just better quality team compared to the opponent.
4. The player is able to manuever a strong start into a "3v1" (even with some damaged mons) and close the game out. This is not always possible but outside of just straight up sweeping the opponent its usually the best and safest path forward for a typical team.

On the other hand a typical factory loss probably goes like:
1. An early disadvantage (usually, the opponents lead just causes way too many problems) that the player cannot recover from.
2. Running into a set or sets that in some way heavily counter or punish the player's team.
3. A close game where the player simply gets "sufficiently unlucky". This could be a stone edge miss, or it could be the AI landing a 3 turn sleep or double full paralysis.
4. The player makes a horrendous misplay. We'll speak no more of this and pretend through practice and improvement we eliminate it from happening again.

What is rare in the factory and important to keep in mind:
1. Full PP stalls. You very rarely ever get the chance to do this and its usually a few specific sets that the AI plays badly that will allow it.
2. Setup sweeps. Swords dance, Calm Mind, Dragon dance arent that rare but safely getting time to setup and go is.

Please internalise that. It's very different to both PvP and ingame play.

The full decomp of AI logic is out on pret, but let me give a brief overview of how it works (this applies to 28+ in singles):
1. The strongest routine the AI has is the infamous "go for the kill if you see it". This is the most important thing to keep in mind and also the most important part of optimising drafts (to avoid giving the AI these freely available strong turns). The second the AI does not see a kill, a lot of its sets can behave semi-randomly and in general I would say all longer matches should be biased in favor of the player.
2. A lot of moves (hyper beam, superpower, explosion) break the above logic and instead operate solely on HP thresholds. These are the most important exceptions to learn as they come up incredibly often and massively transform how deadly some opposing sets are in a functional sense. It's also the most important part of AI manip you can do to avoid moves like these being selected - particularly in doubles.
3. In limited knowledge situations (a classic one is illumate vs volt absorb lanturn, or heatproof/levitate bronzong) the AI will simply guess.
4. The AI's behavior on moves it thinks are useless is incredibly predictable. It will never twave a PRZ pokemon, or hypnosis a sleeping pokemon unless basically no other move is a viable selection. This also transforms how deadly some sets are (sleep + dream eater being the biggest example as these go from being super deadly if controlled by a human opponent to borderline useless in the hands of the AI)
5. The switch-in (2nd pokemon you will see in singles) is solely based on the calculation of what the 2 pokemon in the back would do as their strongest move, if the move was used by the pokemon which just fainted. As a consequence of this, a pokemon with explosion will almost always come in 2nd. Meanwhile a lot of the stall sets with weak or no damaging moves are almost always 3rd. Similarly this can tell you information, such as if [pokemon without ice beam] comes in against your salamence, then it means the 3rd pokemon cannot have a damaging ice move.
6. Non-faint switch outs are very rare, similar to gen 3, and are almost exclusively "resist/ability immunity" switches.

As a consequence of the above observations I'll give a brief list of "sensible" common motifs in games (note that these are not hard rules, please break them in specific scenarios):
1. If your lead pokemon wins the first 1v1 and you don't OHKO the switch in, heavily consider saving the lead even if this means something in the back takes a chunk. Your lead is likely to have a better matchup into the 3rd pokemon because of point #5 above.
2. Whenever possible, if multiple moves will KO an opponent, click a different move to the previous turn. This allows you to smack the rare but relevant cases where the AI tries a resist/immunity switch in regular play.
3. You very, very rarely want to go for a stall angle with the entire team. The most I would do is to take a "special tank" and "physical tank" but I would not push it any further unless in exceptional circumstances and definitely would be even rarer after battle 49.
4. You really want to have something fast in the backline so long as the set is usable.
5. Safety planning - both in drafts and in play - is usually worth it. Thunder wave, Explosion, Destiny bond are all excellent moves.
6. I would say a conservative estimate for the number of "really deadly things" is around 20% of the factory post-49 but also "nearly useless" is also around 20%. Use this to guide decision making in games where you either feel like you are very ahead or very behind.
7. I would say that for typical teams, the % of nightmare fuel (sets that definitely threaten to win a 1v2 with the AI piloting) is closer to 3-4%. That -16% scrubdown is roughly what you gain by intelligently combining 3 pokemon.
I'm aware of Glen and A's tierlists already on this forum.

I think A's tierlist for lower-tiered pokemon is something that largely agrees with my own.
I think Glen's - while funny - is quite wrong and sometimes critically wrong.

https://docs.google.com/spreadsheets/d/14tdALdEvdrDWJ0TXyU04Nr95xuNlWrvmfUE-BM5ohnk/edit?usp=sharing

It doesn't really make sense to wrap everything that matters about a set into a single metric but I also know that the simplicity of trying to boil it down is what people who use guides want.

Brief disclaimer:
- The "tier descriptions" can be kind-of ignored. It's a small attempt to slightly delinearize the process.
- "pick the 3 highest tiered sets" isn't optimal as a draft strategy. I hope that's obvious. It's definitely not what i do. However, if you have no clue or want some guidance or a second opinion on a close call, this is the resource. Just please don't blindly rely on it.
- I have some thousand hours playing with drafted sets in the factory and very similar formats. This doesn't make me "correct". I'm aware of my slight biases on certain things and have tried to be objective as possible here.
- The tiers are specifically weighted to spread sets out. "average" and "bad" should fit the definitions
- Earlier sets (used in rounds vs weak AI) are of course shifted to the "good" side of average. And tbh the tiering for some of the worst sets is occassionally exaggerated to pull things out
- The tiers for "350+" (round 4+) are all wrapped into one. This is because it most effectively captures the comparison between sets/ tiers without me having to tier each pokemon like 3 times for slightly different contexts - and I'm not entirely sure how much this would even change the tierings.

You may disagree on some tiering. In fact i'd be surprised if anyone exactly agreed with this, that would be weird.
A dramatic (multiple tiers) disagreement you may have is probably one of the following:
- I typoed or copied something across wrong. I've tried to sort 950 sets and this list has gone through a few revisions, so a mistake seems likely
- Information has been lost as I've tried to compress "viability" into a single metric (aka maybe its a set that breaks the concept of a linear tierlist)
- You probably havent actually used the set or (alternatively) have used it way, way more than me.
- Possible its also a difference in "playstyle" or "drafting style" which is fine. Particularly in rounds 3 and 4, it is possible to go a lot slower.

Please - if you are going to complain about it or complain about a team that is "highly tiered" by this list losing - please read the rest of the guide first. And keep in mind that even if the best factory-draft team was taken into the battle tower, it probably struggles to go above 300 wins.

This is mainly just one tierlist at a snapshot in time. By all means make your own.
The factory breaks a lot of conventional logic that applies in-game/ nuzlockes/ kaizo/ whatever and it also breaks a lot of the conventional logic in PvP 6v6 play.

Normal:
the tl;dr here is that normal acts as an amplifier for how good a mon is.
The best normal type sets love the fact that they are normal. It benefits them to reduce a lot of neutral matchups down to a simple case of who has the better numbers.
The worst normal types suffer horrendously as they fail to carve out any kind of niche.
The predictability of (from how the opponents reacts to) most normal types makes them good leads later on.

Fire:
Fire is a terrible type in the factory and in general should be avoided. Water is ridiculously common (the factory is randomly pulling sets from all viable options and there are a lot of waters to pull from).
A stupid number of other sets also get coverage for fires in EQ, stone edge, rock slide, etc.
A lot of fire's advantages from competitive play are gone here (steel in particular is rare and most steel types suck in the factory anyway). It's ingame advantage is also gone because the opponents in the factory will run max EVs and match the IVs, meaning you'll probably be disappointed at how little damage is being dealt.
Fire is not _unpickable_ though and the good sets do benefit a lot from how fire types basically always get a real STAB option.

fighting:
Fighting is a mixed bag but ultimately should also be avoided as a type. Although normal types are common, flying types are also common. Hitting Rocks and Steels isn't as good as it may seem since they (largely) suck.
That said of course some of the best pokemon happen to be fighting and fighting does have a lot more going for it for relevant offensive coverage than fire. Fighting types also often get real STAB, which is usually a ticket to avoid being really bad.

water:
The type you should fear the most and also probably the type you will draft most often.
Water types are blessed since they're often surprisingly quick in the context of the factory meta, and often have good enough bulk along with a blessed set of coverage options.
Water is a type you can double pick in drafts quite easily, especially since the primary fear is usually just electric as the majority of grass sets are terrible and slow and can be smacked by an ice beam anyway.
Most bulky waters dont even drop to STAB tbolt. Even something not known for being bulky like Golduck can sometimes take a hit for Jolteon.

flying:
There's no doubt that drafting something immune to EQ is always a bonus to a team.
Flying types being fast also makes them inherently fine pickups.
The thing to be aware of is that flying type itself doesn't help much. Strong grass and bug sets are rare and while hitting fighting is nice, that's about it. A lot of types offer a lot more offensively than that.
Outside of ground immunity most flying types would rather give it up for neutral incoming damage since so many things randomly run rock/ice/electric coverage.

grass:
On paper grass is a type that should work, as an EQ-resist switch with a theoretically good matchup into water should work.
However instead, grass is comfortably one of the worst types in the factory. Seeing "opponent specialises in grass type" is often a blessing.
There are a few to watch out for of course, and you should respect Leafeon, Roserade, spore Breloom, Venusaur and a few others. Sleep powder too. But you probably dont want to draft that.
You typically automatically counter this type with any real team accidentally. it's rarely a concern.
Taking a grass type is often a mistake if its just to cover water, but the strong sets (those mentioned above/ in higher tiers on the tierlist) are fine.

poison:
Poison is surprisingly not that bad as a standalone type. The issue is that poison sets themselves often suck.
Preparing for poison is something you'll automatically be doing while trying to build a good team.
The good poison sets are probably going to surprise you with how good they are. Drapion and Toxicroak both get some bangers. Very usable with an EQ immunity. Do not stress about psychic, its rare to build a team weak to it.

electric:
Electrics often work ok as role compression via their speed and good defensive coverage profile. The predictability of incoming STAB makes them good leads.
Electrics without any real coverage are often too bad to take or keep for long though.
Beware of physical electrics which as a category are largely trash.
Also beware taking something unexpectedly frail here. If you dont know what that means, check if it can live a (neutral) 252 machamp cross chop or 252 milotic hydro pump.

ground:
STAB EQ is a good drug.
But be careful either relying or taking multiple ground types, water is common and punishes it hard, and often any kind of EQ-resist/immune opponent causes too much of a headache.
The frustration of ground is that its best sets are very good in a vacuum but the defensive synergy to get them working can be tough
Beware the sub-60 speed line too. Marowak, Rhydon, Golem can be picked but be very aware of how they're pretty much never going first - meaning they suck at revenging and will have to tank multiple hits if put in to check a set.

psychic:
Psychic sets are largely rubbish in gen 4 and this is an important distinction to make compared to gen 3.
The primary blame for this is the way EV spreads work, most psychics are horribly physically frail; some may as well be weak to all physical moves.
Physics also arent quite as fast as they need to be and too many of them are modest.
That said, there are some good filler and stopgap sets in here. Beyond the obvious Latis, slowbro/king can work, espeon/alakazam can work as a role compressed fast answer.

rock:
Rocks are often horribly slow and way to open to getting smacked by common coverage. Doesn't help that at best you're probably looking for rock slide or stone edge as moves you may have to click.
The exceptions (fast rock types or ones with blessed secondary typing) do work very well though and they're also painfully scary to run into.

ice:
The worst thing about this is how likely they are to fire off 10% frz STAB moves. It actually makes them a problem to face and its rare to have counterplay against that.
The majority of ice types suck of course, but the good ones are great. Just make sure the ice STAB is a good move and not avalanche or ice fang.
ice is up there as the worst type to take multiples of, although its sometimes necessary. Because of this you'd rather run 0 ice types though (so that if forced, then 0+1 becomes 1 and its not 1+1 becoming 2).

bug:
a rare type in the factory and not one to be afraid of. Your teambuilding will automatically skew towards pokemon with coverage and defensive switch ins here.
Heracross and pinsir are usually unwelcome but a lot of the rest either suck or are played so poorly by the AI that they might as well suck.
Bug is an ok type to draft so long as the set is good. The flying weakness is annoying but you will probably already have something with either tbolt, ice beam or stone edge.

dragon:
Generally pretty broken. Easy to find entry points when you know enemy sets, defensive synergy as a switch-in against the predictable ai is good.
The weakness to ice is often either fake or overblown. A lot of ice types suck and non-stab ice is rarely doing enough unless the thing using it has a reptuation for being offensive.
The scarcity of steels also makes these even better.
The downside? A lot of the sets are kind-of bad offensively in a way that will make you overestimate them.

ghost:
A bit like normal this is often feast or famine in terms of making the set a lot better or a lot worse. The defensive opportunities often sound better than they are in my opinion.
It's unusual to get a draft which can take advantage of the amazing ghost immunities but when you do that's absolutely cracked. do not try to force this though.
Drafting ghost sets is basically never an issue though as you will almost never take multiple and the synergy overlaps are often terrible types also, so you need to be less concerned about being forced there.

dark:
Surprisingly good type once you ignore the terrible sets, probably better than new players may expect. Bug coverage is rare in the factory and although fighting is not, teams often can automatically handle these opponents.
The downside is certainly the sets themselves though. A lot run terrible moves like night slash as their only real attack which is close to not having STAB.
They're also often a bit too squishy to actually remain on team as you optimise throughout a round.

steel:
Steel, when it works, is even more broken in the factory that you are thinking.
Fire and fighting types are often easily handled by natural teambuilding, so are EQ switch options. in fact baiting these moves out is often a feature rather than a bug.
The downside? So many steel sets are downright awful. This is both good and bad. It reduces how scared you should be of the type massively (perhaps the biggest adjustment a competitive player should make).
Take the good steel sets when offered though. You won't see them often.
Leads are the most important sub point on all drafts. The lead matchup is frequently what transforms a game from easily won into easily lost.

The question of how to optimally choose leads is - as far as i can tell - barely something that has been discussed at all despite how ridiculously important it is, but I'll give insight here on how I do it at least. See my writeup or videos for actual examples.

I have a few guiding ideas:
1. The lead wants to have as few matchups it would want to switch out from as possible, even if many of the ones it would stay in for are "losing". Momentum is a painful thing to lose and frequently the backline cant be relied upon to fix a "turn 0" problem. Sets with many bad matchups can be viable but they're often non-functional in the lead position as they put such a huge burden on the team.
2. Good leads are usually fast. This is both because fast dominant matchups result in straight up 3v2s but perhaps more importantly because a fast lead pokemon at least will do something in a bad matchup. As an example; crobat4 isn't a great pokemon, but if I can at least get off a brave bird in a bad matchup then this might be enough to safely recover to a 2v2. A huge improvement from e.g. Golem getting smacked by surf and you're now in a 2v3.
3. Ideally in a good team, the lead is the strongest pokemon. This is simply because I want to maximise the time its contact time and increase the equity on these decisions (because its a strong set those decisions should be more favorable more often), but also because often the strongest set is the set most likely to put me ahead and "closing" a game that you are winning is a lot easier to optimise and navigate than being forced to play from behind.
4. The lead should have a typing which makes likely "strongest damaging" moves against it somewhat predictable. This does not necessarily mean the typing must be good, but it should avoid a lot of the randomness that happens from unknown sets appearing at 49+. As an example; electric is pretty good because its either attracting ground moves or STAB moves. Electric's lack of resistances is actually beneficial for planning. On the other hand Psychic/Flying packs a lot of resistances but unknown sets can be super unpredictable into it - a lot of sets can randomly run rock/ ice/ electric moves and others may just hit it with neutral STAB. This makes a switch out scary or random.
5. There are a handful of supporting moves which work well on leads and this point is probably somewhat obvious to anyone; reflect, thunder wave, trick room, etc. all are better options in the lead slot. This doesnt mean you take those pokemon and put them as leads, it just means that if they are B tier as leads then they might be C tier on the bench.
6. The lead is often the hardest part to fix in a average/bad draft. It's hard to have a set rule here but you need to be very aware of what kind of structure you may be locking yourself into. For example Magnezone can be an ok lead but you'll desperately want an EQ immune/resist - and if the initial draft doesnt offer that then you may decide that Magnezone introduces too much risk to put in the lead slot. Similarly if you have a bad draft and shove your best pokemon in the lead, you will be more limited in potential swaps to pick up pokemon which only really work well in the lead.
7. There are a subcategory of sets that I just really like in the lead slot. This includes but is not limited to; destiny bond, focus sash, choice items, slaking, plus anything with a semi-reliable "scout" (protect/fake out/etc ). Unlike point #5 above, this actually would be a tiebreaker for me when deciding between leads.
8. You do not want to over-burden the lead with too many roles. The worst sin (it's ridiculously common) is to have your only fast set be the lead (in which case you might feel you HAVE to save it too often) or to have your lead be the only compliment to your pokemon in the back (e.g. both pokemon in the back are weak to electric or flying but the lead can beat these types - creating scenarios where you feel pressured to save the lead frequently and start kind of 1/2 a turn behind).

Knowing how to balance this shit is hard though and takes serious thought and reflection on all rounds you've played (if they went well or badly). It's easiest for an exceptionally strong draft, where you pick the best lead for the team you're looking at and you're probably done. On a bad/ mid draft; its super complicated. You will need to feel out how likely you are to swap out various sets, what future (unknown) versions of the team could be and what lead they would want. It's difficult to write a guide on that but I discuss it extensively in live comm and my writeups.

This is all somewhat counterbalanced by the simple fact that you do want to pick a good set in general. While it is true that e.g. Electrode4 is a good lead, its also true that its not an amazing set overall and you'd be making a mistake to try and maintain or select it ahead of a good set that doesnt work quite as well in a lead slot such as Heatran1 or Gallade2.
One aspect of drafting that is massively overrated is the idea that the team comp has to matter; or that you may get bonus points or any kind of magic protection for running a Fire/ Water/ Grass core. I see this all the time in challenge runs, factory write ups and just in the general community.

Set power variance is way higher and matters way more than the collective "coverage" of the team.

Lets take a token example; say we want to draft Rapidash-1 (fire blast, sunny day, solarbeam w/ resto chesto) , Feraligatr-2 (4 physical moves with stab), and vileplume-3 (3 attacks + SD). I have intentionally not picked the absolute worst sets here.
Is this team "good"? On a very brief surface level i think some people might say "yes". It probably passes the "level 1" analysis (forgive my caricature of what i interpret the thought process to be for some players):
"There is good type synergy"
"There is mixed category coverage" (both physical and special)
"There is a fast lead pokemon"
It also has a small "bonus" team synergy where Rapidash can set up sun for Vileplume.

I won't call such a team terrible. But it is well below average, and only something i'd expect to see on battle 1 or 2 of 7. I'd say such a team is maybe 70-80% to win a single game against the AI.
Rapidash-1 perhaps deserves the label of bad. Generically speaking, we're hoping to win the opening matchup with 2 fire blasts and we're hoping the opponent cant ohko. If we're lucky, we can OHKO something weak to fire turn 1.
A few problems immediately pop up though:
1. hitting 2 fire blasts is only ~72%. phrased differently: a 28% chance that rapidash probably loses a 1v1 to just some generic opponent.
2. Rapidash has exactly 1 immediate damage move. Fire is good complimentary coverage, but bad solo. There are a _lot_ of water types in the tiers 4-8. There are also a surprising number of rock types. You may think "i can just sunny day solarbeam" - but this means you faint to any kind of OHKO response without doing anything, and also Rapidash just lacks the raw power to OHKO a lot of things with super effective solarbeam.
3. The pokemon we OHKO with fire blast do not include many primary deadly threats. In factory we stress about facing pokemon like tauros, latios, snorlax, flygon, starmie. Its nice that rapidash can hit say, magnezone or scizor. But that's a pretty small overall contribution in the context of ~600 potential leads
4. As alluded to, rapidash lacks the bulk to actually hang around against anything "kind of" strong. Jolly Aero secures an OHKO against it, and so do pokemon that probably dont immediately enter your head as threats during drafting, such as Blastoise3 or miltank4.
5. The pokemon we can actually use resto chesto to buy an entire turn against is a short list and again full of sets that are kind of handled by "anything decent".

A bad lead is a serious issue on its own. In this team its even worse because both mons on the bench are slow. If rapidash loses the opening exchange then the switch in is taking a hit.
Even if we can be "smart" about the switch in, it's not like vileplume appreciates being hit by a move like surf. In fact given the AI prefers to switch in a pokemon with the strongest base power move into yours, that's probably vileplume dead next turn.

Under more examination even more holes appear. Our famed Fire/Water/Grass core suddenly starts looking a bit shaky. We wanted to switch out against something like aero, but aero can beat feraligatr after getting a free hit in.
If we're expecting an earthquake, vileplume doesnt want to take that either (even if i'm generous and magically remove the poison type temporarily, something like swampert1 can still 2hko vileplume3), and against a faster opponent it probably just outright dies on the switch in.
If we're expecting a water move, unfortunately water/ice is a really common offensive combo among factory sets.

The team also lacks the ability to close or trade up. Once this team is behind, it loses to a lot of simply average sets from the opponent.
The way such a team wins pretty much requires rapidash securing a KO, getting another hit in, and then using the free switch to create a 2v1 or a 1.5v1, and then hoping the final mon isnt something we wanted rapidash for.
What actually saves this team from being unworkable is Feraligatr2 being a legitimately good set. But Feraligatr2 is a set that can work on a number of other teams, there's no need to force it here.

Again, I intentionally wanted to choose sets which seem ok here. These are not the worst fire/ water/ grass sets available in the factory. And also in fairness; this team could take a win. There are worse drafts. But a draft like this is something you would want to escape almost immediately.


To take an opposite approach; what about a team like Garchomp1/ Salamence3/ Flygon 1? Is this team bad?

The kneejerk reaction, particularly coming from 6v6 PvP metas where people specifically prepare for common threats, is that such a team is shaky. Surely this team is too weak to Weavile, Choice scarf users, Starmie?

I'm not going to call this team amazing. In fact you'd probably want to replace at least 1 pokemon if something acceptable comes up. But this is a strong team by factory standards and probably has 90%+ winrate in a single game.

"What about ice pokemon"
Its important to remember that to lose to an ice type, you have to get hit by its ice moves. Glalie wont even hit 2 members of this team before it faints. Jynx wont hit 1 (most likely). Mamoswine will drop to mence or at most need a single EQ assist before/ after. If there are multiple ice types, we'll be told before the game.
"What about good ice pokemon"
There are some things that may feel like a bigger concern. It is true that Froslass1 just wins here. Adamant weavile can win too (jolly will struggle with some rolls after intimidate).
So is this bad? we're talking about 100% losing instantly to weavile1 (probably weavile4 too), froslass1 and probably froslass3/4 (although neither would be guaranteed alone). lets round up and call it 5 sets. Lets also assume that as this is a quick post that there are 10 other sets that outspeed and OHKO all 3. That's still only 15 sets out of 600.
Although the weakness is more obvious, this is a much shorter list than what would instantly threaten and "effectively" beat the fire/ water/ grass team above. Better to lose 100% of the time to those 15 sets than to lose 80% of the time to 150+ sets out of the 600.

As for other "OU kneejerk responses"; the factory lacks a lot of them. Scarf is a very rare item. Only 1 starmie set actually runs ice beam (maybe the blizzard one hits all 3, if it doesnt its in trouble), and the ice beam set is modest so garchomp has a small but not insignificant chance against it.

This team also has the benefit of just having 3 pokemon that are not only fast but can realistically 1v3 or at least 1v2 + another hit a decent number of sets.

Again, this is not a perfect team. But we probably only need to run it for a single battle before the "ice problem" can be fixed with a single swap. The reward after that would be a team that most likely stomps the remaining matches.

I cannot really think of a scenario where you'd intentionally take 3 dragons (perhaps only and exactly if its the first match and the other 3 mons are truly unpickable), and so this is more of a hypothetical.

On the topic of teamcomp, while its fine to be weaker to things (like being weak to exactly fast pokemon with ice beam, or being weak to exactly fast fire types that dont cripple themselves as they do damage) - its not fine to be weaker to others.
In my opinion it's basically completely unacceptable to be weak to water. I'd also say having nothing that is physically tanky at all is very dangerous and can only work with an exceptionally fast team.

It's also pretty bad if the entire team lacks any real coverage. This would be a scenario where somehow (because bad sets were taken) perhaps the only coverage moves are grass, poison, normal, bug. And then maybe you just straight up lose to some steel or flying type.

Final disclaimer if it isnt obvious:

The absolute best teams you can draft do have good team comp. For example; Salamence/ Magnezone/ Snorlax is good and a significant part of its power does come from type synergies. This section is about challenging the idea that team comp is a high priority, or even a priority at all in most drafts.
(cont. in next post)
 
Last edited:
10 questions for is this a good factory pokemon? / Where does the tierlist come from?

0. The "quick" comparisons
With a damage calculator or otherwise check how many of the following pokemon the set can beat 1v1:
Wailord1, Froslass2, Espeon1, Miltank1, Donphan2, Feraligatr3, Drapion3, Luxray3, Glalie3, Rapidash4
This is a decent test to give you an idea if the set belongs roughly above or below the "average" line.
None of the sets listed there are strong; in fact they're all about average by 49+ factory standards.

--- Trying to promote my mental calc model but also aware that in the era of damage calcs and "1vAll" / "Allv1" then maybe you can ignore the next 2 if you're always going to be more willing to open up the website than multiply 2 rounded numbers together.

1. What is the raw offense power of this set? (Include STAB) Below is a lvl50 discussion but the same principles apply to lvl100 except the numbers will be larger.
examples (for generalized purposes you can drop the 3rd significant figure):
Garchomp4: 200 atk, 120 bp outrage -> 180 w/ STAB. That's 36k.
Latios4: 200spat, 140 bp draco -> 210 w/ stab. that's 42k.
Empoleon1: 150 atk, 120 stab waterfall. That's 18k.
Raikou2: 140 atk, 100 bp return/ STAB tfang -> 14k.
aero2: 150 atk, 65 bp tfang -> thats a bit below 10k.
To give some context to these numbers, 30k is about the benchmark where you start to outright OHKO the frailer stuff in the factory (without any kind of super effective modifier). This builds until around 50k where even things considered either physically or specially bulky start to drop.
20k-30k is pretty much the range where more and more things stop avoiding the 2hko.
below 20k you are looking at a pokemon that will not only struggle to 2hko "the average" set, but will also struggle to OHKO in a lot of its good matchups.
Below 15k we're talking about actively disappointing damage that starts to miss out on even a 3hko against averagely bulky sets. At this point without super effective coverage way too many matchups start to be mathematically lost.
Below 10k we're talking about sets that dont really have an attack. Aero2 will miss the 2hko on most water types with thunder fang. Anything in this range better justify itself with its non-damaging moves.

There is of course an additional layer here which is how powerful the next strongest move is. This is a metric where Garchomp, Infernape, Swampert and other "known" good pokemon tend to excel. Having said that, I'd favor a set with a single huge attack over one with multiple weak attacks.

2. Whats the raw bulk of this set? (changes post 49):
-> pre 49: can uniquely consider just 1 defense stat. You will know the enemy set on sight. This means pokemon like Skarmory and Blissey can work well, as the majority of sets stick to either physical or special alone.
-> post 49: must consider both or the weaker of the 2.
Median bulk is something around 18-19k and this is probably slightly higher than you are expecting. A lot of "bad" sets (things that slip your mind such as lickilicky, exeggutor, cradily) are bad because they're slow and slightly bulky, but these sets drag up the median.
10k or lower is embarrassingly frail and sets below here are genuinely concerned about fainting to a single STAB hit. Absol is around here. Some terrible sets like Mr. Mime and Jynx are below 10k on the physical side. Dugtrio is well below this. Through the lens of factory hax, pokemon in this tier can never be "good".
15k (e.g. 150 hp with 100 physical or special defense aka around base 75/80/80 with no EV investment) is roughly the line that when crossed exits frail territory. A 15k bulk score is likely to be 2hko by a 15k offense score. That's OK so long as you're faster or in a favorable type matchup.
20k (e.g. 170 hp with 120 def) is a bulky set, at this point the 2HKO requirement is more like a 25k offense score (for context, modest raikou tbolt is that ballpark). Pokemon crossing here will suddenly start winning a lot more 1v1s even if they are slow, but its not the level of bulk where you can freely switch in on a generic threat and be ok.
25k or so is the point where not only could a set enter into a STAB move but then also pretty reliably stay in and potentially take another 2. This is also roughly the ballpark where few attacks could OHKO.
Above 30k is where walling is actually a real thing in the factory. Above that point with a 50% recovery move you can actually stall a super effective stab move on that end of the spectrum if the set you're looking at isnt known for being an offensive monster. This is why Cresselia and Hippowdon (physically) consistently actually work. Still, be careful as the chance of 1/10 moves critting is "higher than you think".
If you want context on just how fat 255/255 blissey is, it scores well above 50k. Both (2 and 4) the 255/255 sets have softboiled and therefore basically cannot lose to a special attacker. Physically speaking, 0 def blissey is around 11k which is just enough to not drop to most strong STAB moves.

To pre-empty any curiousity here; in the generalised abstract case, it doesnt matter how stats are distributed to achieve the overall bulk numbers. For 25k bulk, a 250hp/100def set will have similar XHKO profile to a 100hp/250def set. It matters *slightly* more at level 50 as damaging rounding isnt that favorable as defense/ spdef stats get huge but you can almost entirely ignore this detail.

Even if you are familiar with how bulky you think things are, this is a good objective reality check. Whiscash is a pokemon that perhaps has an undeserved reputation for being tanky when objectively speaking its only around 20k and simply bulky. It has good typing still, but this demonstrates why it does not work as a check or a pivot generically. As a real example; 252hp whiscash isnt coming in against dragonite 1's dragon claw and always avoiding the 2hko.
A lot of grass, electric and fire types may need an additional check here as when I call these sets dubious its often because of sub-15k bulk on a stat. Magmortar and Manectric would be much better if they had around 20-25% more bulk.
Some things may slightly exceed your intuition also. Blastoise is considerably above 20k, so is meganium. This doesnt make them "good" on its but its important to properly have an understanding on what you could expect them to do.

Of course all bulk is filtered through the type lens. I guess I trust anyone reading this to already have a good intuition on what types are good/ bad defensively, but its important to remember that while e.g. Lucario does have great typing and can use it to enter, it's still effectively sub-30k even through a not very effective multiplier. Anyone who has used lucario in any context already knows this of course.

You don't have to do any of this, particularly if you aren't making a tier list. It's definitely enough to go "yep jynx is shit because it can faint to a strong STAB aerial ace".
I'm just being as transparent as possible about how I've always thought about the game since I first found a serebii.net page on how damage works god knows how long ago. Not having to focus on this and just knowing the values for various pokemon helps me shortcut a lot and focus on more important things in drafting and play.

40% (80% at level 100) of atk*power/def is also something I do a lot still. This is a quick calculation, perhaps a lot quicker than people realise, and the more you do it the easier/faster it becomes. It's rare to actually care about exact rolls, particularly outside of a specific 1v1, and when analysing / judging sets this is a more useful metric.

--- mental calc promotion over

3. Can this pokemon function as a defensive pivot and to what extent?
This is really important because if it can it often gives you a "5th move" on both of your other pokemon.
The AI is very predictable and this is the biggest advantage you can have over it alongside the additional option to switch
Examples of things that massively benefit or promote pivoting:
1. Being fast.
2. Good typing but particular emphasis on dragon, steel, ghost, flying, ground
3. Useful pivot ability. Water absorb, Intimidate, levitate, natural cure,
4. being as bulky as possible (see above)
The annoyance of defensive pivoting is that its often a snowball effect. Once 2 pokemon on a team support it, suddenly the job requirement for a 3rd member to work goes down massively. But this is why you cannot force it in drafting.

4. Is this pokemon fast?
Speed is the most important stat in the game. A slow pokemon can be good but can't be incredible.
Being faster is often what saves matches and allows a pokemon to 1v2 or contribute significantly later.
Pay particular attention even to slower sets. It's easy to intuit how 155 speed is a huge improvement over 104 (numbers chosen intentionally yes). Hidden in this; 85 is massively faster than 60. There's a lot of slow shit in that speed sandwich, and a lot of that shit might matter.

5. can this pokemon dominate other sets? (win the 1v1 while taking almost no damage at all or while "winning the game early")
Particular emphasis here is on the ability to outspeed + OHKO, or the ability to outheal (recover, rest, etc)
example: snorlax2 isn't a great set but it does dominate a lot of special attackers which is useful.
Golem4 is ok but dominates almost nothing. its going to be hit by a coverage move even if it can win.
Also relevant: dragon dance/ calm mind/ curse/ bulk up on fast/ sets with healing can setup wins.
Certain types and type combinations are better at this than you may think. Steel is probably the easiest example; a shocking number of sets do not run coverage for steel (in the sense that all their moves may be not very effective). Remember, hidden power does not exist around these parts.

6. What is the *safe* raw offense power of this set?
very often in factory the opponent will be around 20-40% HP.
While its fine to occassionally click 80-95% accurate moves, you really want to avoid this as much as possible
sets that can safely finish off the opponent with near certainty can boost the team's reliability a lot, but also benefit from it themselves. It helps a lot if after landing your inaccurate move that you can pivot to aerial ace or fire punch or whatever.

7. can this pokemon close 3v2 or 2v1 scenarios?
A very common scenario in the factory is to be winning 3v2 after the initial 1v1 (either via a switch or not).
An incredibly useful tool is therefore the ability to safely close these games.
Main focus: explosion, self-destruct, destiny bond.
To a lesser extent: simply being fast and either hitting hard, or being able to safely criple (example: thunder wave) work here

8. can this set be clutch in 2v3 or 1v2 (or worse) scenarios?
Another common situation is where the AI randomly gets a great lead matchup against you.
e.g. zapdos vs gyarados lead.
In this situation even if you save gyarados by sacking something else, the team has to recover the tempo, how effective can this pokemon be at that?
Speed is very important here - so are some riskier filler moves that you may not clicks as often (setup moves, etc).
Some luck based moves matter a lot also. Confuse ray is not a move you usually want to use but it can be excellent here. same goes for hypnosis, flinch moves, etc.
some of the endure/reversal/salac sets are amazing at this too.
focus sash, counter and mirror coat can be really useful tools in preventing the AI from rampaging through the team. These also lessen the burden on having a trio of pokemon that can handle "everything" because so much is handled via them.
destiny bond and explosion once again shine here in terms of rescuing a bad opening matchup.

9. How limited is the set in terms of being "useless" into a large chunk of sets?
This can place a burden on the other team members and on the way you have to play.
Examples:
Electrode2 does almost nothing if the opponent resists/ is immune to electric
umbreon2 does nothing to toxic immune pokemon
sceptile2 loses really hard to a lot of sets while (usually) achieving very little
Anti-example; counter blissey can get counter off against a surprising number of physical opponents, although its role in this is probably limited to being already in. Blissey is threfore not useless that often.

10. a note on the set's gimmick, if applicable:
(Note that I will call everything that isnt 2-3 regular coverage moves a gimmick. I am fully aware that outside of the factory there's nothing unusual about a toxic stall umbreon)
main question is how luck reliant the gimmick is, against how strong it is.
examples:
ambipom2's gimmick does not require luck and is ok if your playstyle involves almost never switching, although a bit limiting in forcing ambipom to be played in a certain way. This makes it low risk/ low reward.
milotic2 (among others) relies on huge luck. its basically just an evasion sweep attempt. Even if these work most of the time, the AI will eventually just not get unlucky. If/When this happens you're looking at a 3v2.
Less relevant for gimmick sets is how they work on a team;
The fact that e.g. sunny day doesnt pair well with water types isnt necessarily a knock against the gimmick set. Of course in a real drafting scenario you have to think about it. But when judged in isolation it can be ignored.
Trick Room probably less so. Trick Room is a devastatingly distruptive move in terms of how it flips the tier list. The gimmick here extends beyond just the set. The issue is that it's difficult to build a trick room team in the slow drafting process, and this makes Trick Room worse than you may think.
10 Common pitfalls in set analysis in rough order of importance/relevance, or, 10 things you don't make a snap decision on:

1. "This set is bad because it's using the weaker primary offensive stat"
This is maybe the biggest overreaction that players have. You must isolate the set. Forget about its unused attack stat and look at what remains.
Garchomp2 is a great example. Oh no, we're not using 130 base attack? But wait... dual unresisted stab, 2 strong coverage moves, amazing speed, very effective bulk. Yes, this is probably a top 10% pokemon. It also (hidden buff) prevents Garchomp1 or 4 from showing up as opponents. That opponent-Garchomp prevention thing is a joke, but also not really a joke.
To continue with another dragon: Latias2. Oh dear, adamant latias? But wait, its still base 80 attack which isnt terrible, plus its faster than most of the factory, plus it has a very useful set of resistances and defenses, plus it also has dual stab, EQ and a move for DT spam? It's the worst latias set. but its probably top 30% of sets.
Beware the times when this does apply though: Manectric3, Dugtrio2, slowbro2 - these are bad sets. Even Raikou2 is pretty crippled by being physical, although at least Raikou2 is still a fast and not-frail set.

2. "This set is bad because it doesn't run STAB"
I will not deny that STAB-less sets are worse than they "should" be and often would be improved a lot. But it's really important to rip the set out of a vacuum and consider in isolation where it stands to other options.
Tyranitar2 and Tyranitar3 are going to be my examples here. In my opinion these are top 33% sets and deserve respect both to draft but also to face. No STAB is a loss but Ttar is kind of a "660bst" mon in gen 4, and a lot of its pitfalls in OU dont apply here since the enemy team is random.
Secondary callout: Regirock2 is also easily in the top half of factory sets. It does _look_ bad and you will need to run anti-water on a team with it, but its a hell of a lot better than an actually bad set and this thing really does chunk attacks.
At the same time, is Aerodactyl2 bad? Yes. Yes it is.
What about Registeel3? This is actually ok, not good but not bad. The privilege of being stupidly bulky with good typing is real.
It's on you to recognise the difference between the examples above and why some work and some don't.

3. "I recognise this mon from OU/UU! It's good"
Probably more of a thing for newer factory players. There are certain pokemon that work in OU mostly due to one of:
- the frequent switching in singles PvP (The factory AI almost never switches)
- the fact that the singles meta in terms of type %s is wildly different (steel very common in OU, but is a very rare factory type)
- the fact that in 6v6 you can have a more specific/narrow task or role
- its OU/UU competitive set relies on something you may have overlooked the importance of. Hidden power is a common one here, so is "the right item", as is reliable recovery and (important to check) the right ability. Also just, the importance of having speed EVs...
Dugtrio, Bronzong, Skarmory, Forretress, Blissey, Dusknoir, Roserade - not an exhaustive list at all but a start. The factory version of these sets will disappoint you. Doesn't necessarily make them bad.
Meanwhile Moltres, Cresselia and Tauros are great in the factory. Slaking also.

4. "I'm aware of #3 and I now know best (I'm a hipster)"
I see this a bit too much also. Maybe some people want the factory meta to be magically different and reward enlightened minds. Or those who've played it a lot want there to be some magic secrets.
To bring things back to reality, there is a strong overlap between what people like to use in PvP and what works in the factory.
Speed matters a lot. Bulky waters are really good. Garchomp and Latios are exceptional. Metagross, gliscor, machamp, hippowdon all work really well ... there are fewer surprises than you are hoping for.
Something I hammer home in other areas of this guide rings true; pokemon is mostly a game about numbers. Optimising the factory and PvP teambuilding both acknowledge this boring truth.

5. "This is slightly worse than X so its bad"
Hipsters, this is a point back to you. You get to be right here.
Flygon, Blastoise, Tauros, Golduck, Whiscash, Quagsire, Staraptor (not 2) - these are good sets. In fact some of them are _really_ good.
Do not be distracted by how something is slightly better. This is a drafting format, you're unlikely to see Garchomp, Suicune or Swampert. Take their juniors when offered.

6. "This set has a lot of coverage, so its good"
Be very careful with this one if you haven't thought about it mathematically before. There's a reason why a lot of OU-optimised sets end up looking like they do.
As soon as a pokemon has 90+ base power STAB move plus any additional coverage, there's often little room for relevant coverage beyond that. With STAB EQ, how often does Fire punch matter? There's a list of pokemon which is the answer to this question, but it's probably shorter than you think. In the factory a short list means low probability of mattering.
A lot of sub-70 power coverage moves (looking at the elemental fangs in particular) also fail to matter and this is usually the power range where you are picking the move for text ("drop enemy speed" is definitely magical text, and "cant miss" or "goes first" are also great ones).
The factory likes to throw around elemental coverage in particular and its often a waste of time. Lopunny4 can't pull this off. Luxray4 also somewhat embarrasses itself.

7. "This set has recovery, so its good"
This is one you have to have some experience with factory power levels to really call out.
Recovery (particularly not-rest) is often useless or broken and there's a very fine line between the two.
On Cresselia, moonlight is broken. There are way too many sets that cannot break through this. It only has 5pp but thats enough to clown on an opponent.
Skarmory2 is a complex set for analysis in isolation but the summary is that it definitely can pull off roost.
On Medicham2 and Alakazam2 recover is an incredibly niche move. Medicham2 actually works as a set, but its because you should be clicking drain punch whenever the move isnt resisted. Alakazam takes more than 50% from an 80 power move from a base 90 (attack) invested pokemon. You aren't taking STAB special moves well either.
Special shoutout for leftovers although #9 also goes over it; leftovers is not real healing. It's sometimes a huge problem on the enemy evasion sets. do not try to replicate this against the AI, accept that its something weighted against you due to the asymmetric nature of the AI only having to win once.

8. "This set must be good/ a good lead because its fast"
Speed is the hallmark of stronger sets but on its own it obviously means nothing. There are plenty of sets which are rubbish.
Electrode3, Weavile2, Flygon2 - these are all bad sets. Their speed isnt compensation for this, but instead its merely just enough to avoid putting them in the absolute lowest tiers of the factory.
Be particularly careful of applying this to leads though. A lead's job is not to be fast, its to best engage with the below-average matchups the team could be forced into. Speed helps for this but its more of a secondary effect.
Keep in mind how small a 3v3 team is. Do you actually have room for jolteon? in 6v6 you do. On 3v3 you'll want to use as much role compression as possible. Jolteon is an inflexible pokemon. It's ok. You'd rather be rid of it.
Also seriously consider in bad drafts you may not get the chance to put something fast on the bench for a while. A fast pokemon (particularly one thats also good) may be better off away from the lead position.

9. "The item is good, so the set is good"
With the exception of choice items and focus sash (and a few pinch berries)- we're often talking about a very small change to a set on the players side.
Brightpowder, leftovers, Quick claw, "super effective" berries, lum, etc are often things we are most concerned about when on the opponent's.
Focus sash alone doesnt save a set of course, and neither does Salac/ Liechi/ Petaya. Analyse the main attributes (the numbers) first.

10. "This set seemed good when the AI used it! / This set /almost/ ended one of my runs"
This one especially applies to swapping immediately after.
Do not be results orientated. The thing to remember is that its the players' goal to win a lot of games in a row. The AI only needs 1 win to end a streak.
This leaves a subset of sets that are a nightmare to play against but are not reliable picks;
- Lucario2, Starmie2 and Garchomp3 all definitely belong here. They're unbelievably scary to run into. They're _pretty good_ sets for the player to use too. But they're not top tier.
- OHKO sets unsurprisingly also belong here.
- Evasion spam also no surprise. Skarmory2, Cresselia2, Registeel2 and a few others can work in the players hands. But be aware these are definitely outclassed by a few more conventional sets.
- Sleep is also fickle. Although some sleep sets work, and quite a few of the AI sleep sets can be easily stalled out of all their PP.
A big deal is often made of having both physical and special attackers, as well as never taking multiple pokemon of the same type.
In gen 4 this is a slightly smaller concern and you want to be careful about this rule baiting you into taking worse sets.

There are a few pokemon in the factory that are exceptionally annoying for an all-special team (blissey being the main one - Snorlax/ umbreon are not as bad as gen 3 due to special fighting moves existing as well as special moves in general being quite a bit stronger)
As for type synergies, while an ideal/ very good team is mixed to avoid a loss (or being heavily crippled by) a set like Blissey - its important to remember how rare blissey is compared to "everything else that might end you".
The much more pressing concern is once again "can your pokemon actually 1v1 a good number of sets". It would be madness to take a bad physical attacker over a pretty good special one just because of blissey.
Assuming the 3 special attackers represent a reasonably large range of offensive move types (they usually do; special attackers usually have more coverage options), this is not a bad team. Similarly to the dragon example, you would ideally pick up a strong physical set via a swap of course.

For all-physical teams, the danger is even lower. There are a few sets that can cause issues once again; slaking, Hippowdon and cresselia come to mind.
The reason why 3 physical attackers may be a more common draft has more to do with just how strong and reliable primary STAB options are on the best physical mons in the factory.
STAB on a 90+ power 100% accuracy move without a lot of resistances is unusual and the way base stats are distributed mean more pokemon have a respectable base attack compared to how many function just off special attack.

Outside of mixed attacking considerations among the 3 sets, there are a few other important notes to take on your team's potential for offense:
1. Coverage for water and normal is unusually important. These types are both common and have a lot of strong sets. This doesn't necessarily mean super effective moves. For example, if one/two of your stronger team members has an excellent neutral stab attack, that can be good enough.
2. The total type-spectrum coverage should be good, even if the actual team members themselves dont have great coverage. For example, slowking4 and espeon4 together cover a lot of the typechart. In fact, they kind of cover a lot more than you need.
3. Unless the team has a true S-tier mon, you want any 2/3 of your mons to at least be able to trade with common types. An example of this: a Fire/ Fire/ Electric team can lose very quickly if the electric mon faints and one of the remaining opponents is water/ ground/ dragon/ etc. Positioning for this is a part of having the correct lead.
4. Any kind of coverage you have on a slower pokemon deserves more scrutiny and should be viewed as conditional. To continue an example above, Slowking4 is a decent set. But it's not comfortably switching in and winning 1v1s after a free hit due to how slow it is. Slowking has "revenge" coverage, not active coverage.
5. Conditional attacks should probably not be considered as coverage at all. This is Focus punch, Dream Eater, Sucker punch, etc. In fact by clicking moves like this at all you are admitting that your draft is probably terrible (which is ok, it happens to everyone at some point).
Points on playing against the frontier AI.

1. The golden rule to understand is that the AI's desire to go for the kill outweighs almost everything.
You want to actively stop this algorithm from kicking in as much as possible in my experience. This is why bulky sets are so good. It's also why types like water and steel are so much better than fire or ice.
That said, there isn't a lot of overlap between "bulky" and "fast" without also automatically being an "obviously strong" set.
But there is a magical and massive difference between continuously being in scenarios where the enemy will 2HKO you and scenarios where its an OHKO. The former includes maybe a 40%+ chance of the AI throwing by doing something dumb like using sandstorm, toxic or bulk up. The latter you know whats about to happen and the most you can do is take it or switch.

2. The AI's 2nd pokemon (switch in decision) is a very simple algorithm:
The AI will check of the remaining pokemon that it could switch in, which one of them has the move that would have done the most damage if the pokemon which just fainted had used it against the last pokemon you had in.
e.g. I have aerodactyl and I knock out alakazam. In the back the AI has a waterfall gyarados and a thunderbolt jolteon. Although both waterfall and thunderbolt are super-effective, because alakazam obviously does way more damage with tbolt than waterfall, Jolteon would be sent in.
In fact even if my pokemon faints (recoil, explosion, dbond, whatever) then this algorithm still applies to what i just had in was.
This often leaks what set the 2nd pokemon in is _slightly_ more likely to be, as well as (once known) how much you can narrow down the 3rd set.
In the aero/alakazam/jolteon/gyarados example, not much could be deduced about the 3rd unknown set to me (gyarados) but if the AI had sent in something like ursaring, then I may be able to massively cut down on how many sets are possible in 3rd.

3. Certain moves are very poorly implemented in the early gen AI and are not factored into the "kill" algorithm.
Hyper beam (and equivalents), superpower, solarbeam (and weather), explosion, focus punch, last resort - are all either very predictable or borderline broken in terms of how they work.
This is important because it transforms what should be some very scary opponents into lemons, although still lemons you should respect and carefully plan on how to KO since (most of) these moves get promoted heavily when at certain HP thresholds.

4. The AI very rarely switches, but ..
When it does switch it will almost always be either a resistance switch or an ability-immunity (e.g. volt absorb) switch. For this reason, it's always slightly more optimal to click a move you didn't click last turn (if multiple moves definitely KO).

5. There is no magic retroactive decision making in the AI's thought process.
The AI does not know, nor can it predict that you will switch or that you will/ will not use a move like dive or protect or rain dance.

6. Some move behaviors are very simple once understood and can be heavily abused.
Biggest ones probably include hp-recovery (based on their % and also often based on if they are faster or not), as well as "boosting" behaviors (generally, the AI will stop using any stat-boosting moves at lower HP)
Countercoat is also something I'd underline here as a super predictable enemy behavior once its understood.
Frontier veterans will already be familiar with the AI's obsession with speed control (use twave/ dragon dance/ etc if slower)

7. Understand that the AI's knowledge of what you have is guesswork until it _knows_
Classic example is bronzong. The AI will randomly guess if you do/ do not have levitate until it sees a ground move miss. This is a soft buff to all ability immune sets (the AI may Tbolt Volt absorb lanturn, or indeed it may not tbolt an illuminate one).

8. It's often a pretty big commitment to sac your final fast set.
Mentioned earlier but i'll say it again; there are a somewhat deadly subset of enemy sets that can be very hard to deal with if slower and with limited switch options. Most notorious on this list is probably choice sets but there's also some less immediately obvious stuff like any set with explosion or dbond (these moves unsurprisingly go up massively in pick % once at low HP).

9. Unless you are going super slowly, it's probably a mistake to get tunnel vision on type spec mid-match.
If you see an opponent is an X specialist and pokemon 1 was that type but pokemon 2 was not, I recommend quite heavily against playing fully into that type when there are multiple "reasonable" choices.
The reason for this is that the set variety in the frontier is quite deadly. Keeping a grass pokemon around for the water spec just to see gyarados or tentacruel in last is something that can easily happen. Similarly you may decide to take damage on a bulkier physical set, reasoning that your fast special pokemon should handle the final rock type, only for that to be revealed as aerodactyl.
In my own experience instead of heavily playing into these scenarios, the best "default" play often wins out as the best play anyway. This isnt rocket science though - it's usually trying to maintain a 3v1 instead of a 2v1 or to avoid taking as much unnecessary damage as possible, as well as avoiding over-setting up on a DD/ CM set in case the final opponent is a random QC idiot.
How often you are presented with "multiple reasonable" choices probably depends on how good you are at spotting options, although it's not massively common to have 2 plays that are roughly equivalent in both risk and average outcome.

10. Keep an eye out for scenarios which are trivial PP stalls and scenarios which are not.
Trivial PP stalls are often against a hypnosis/ dream eater set. Allow one pokemon to sleep, then constantly switch between it and a non-sleeping ally. The AI is likely to get stuck on hypnosis/dream eater alternation and you can stall out the set's PP.
Also be aware of when this is not possible though. For example, Gengar2 is totally capable of shooting a shadow ball off.
A few other "danger" scenarios exist that Tower trainers will be aware of. Switch-stalling (or just intimidate switching) is very dangerous against certain variable power moves such as revenge or avalanche which can behave randomly and be chosen even if weaker than another move or resisted by the pokemon currently in.
Overall these stall scenarios are rare and I would in particular recommend caution if its a stall setup where one pokemon is being at all significantly (say, 15 dmg or more per hit) chipped. If you fuck up and enter one of these please do not stay on the sunk cost fallacy boat.

11. Once the AI is out of PP on a move, its behavior can start to seem erratic.
This is because a 0-pp move is still considered viable in a lot of the internal AI algorithms, except it cannot be selected.
Anyone with "slow team" experience as well as a lot of "gimmick teams" (drapion setup, etc) will be familiar with this even if you do not understand why it behaves as it does.

12. understand just how good or bad your draft/ team is and just how good or bad your current battle is
A good team can make safe concessions during games. This means things like using aerial ace to 3hko an opponent with brightpowder, or switching specifically to scout a set later on.
A bad team cannot do this as its margin over random sets will be small. A bad team will be forced to use 70% thunder in some scenarios.
It's important to correctly estimate the team's chances against the last 2 unknown pokemon.
It's also important to experience just how often you actually win 1v2 scenarios. They always feel very scary and almost like a certain loss, but the player's draft advantage of having (usually) better than average sets, combined with just how bad some random sets can be, mean that this is possibly slightly player-favored, particularly if you're smart and kept your faster/ better pokemon for this scenario if you saw it coming.

13. As an addendum to #12, once you know what you are doing in the factory then you are probably like >90% favored in most typical games
This means the best play is probably the one that removes as much variance as possible. So, going for the more accurate move, or choosing not to switch if switching opens you up to a loss if 1/2 moves crit instead of 1/1.
Note on 28-49 vs 50+:
I still view wins 28-42 as a deathtrap, perhaps the hardest point of the factory. It's too easy and too common to get stuck with a useless draft on specifically round 5 and occassionally round 6.
Streaks of 50+ have some advantages despite "unknown" sets you have to scout. Mainly, a lot of sets are actually kind of bad and the variance of it maybe being set 2 often works in your favor. Everything being 31iv is also slightly better for the player as this reduces the effective offensive power of every set (atk iv going up increases offensive power a lot more slowly than both hp and def ivs going up). It's mentioned above but you want games to go slightly longer if you think you're smarter than the AI.
For a long time the wisdom has been to do 28/28 swaps for all 4 elevations in round 5. I give you, the reader, permission to break this. Thorton at 21 is enough of a danger (plus 20 wins is enough time investment) that I do not consider 28 swaps mandatory. I'd rather get to round 5 (28 wins) with ~96% consistency but only with 3 elevations, than get there ~85% of the time with 4 elevations. You should still go for at least 21 swaps total before 28 though.

Thorton:
It's mentioned in other places but:
Thorton1 (i.e. battle 21): in Platinum lvl 50 + HGSS (50+100) this is tier4 (set1) pokemon only, with 12 ivs. In Platinum lvl100, this is tier7+tier8 (set4+legendaries(any set)) pokemon with 12 ivs.
Thorton2 (i.e. battle 49): in platinum lvl 100 + HGSS (50+100) this is tier7+8 (set4+legendaries) pokemon only, with 31 ivs. In Platinum lv50, this is tier4 (set1) pokemon only with 31 ivs.

Basically, Platinum Lvl50 is _slightly_ easier than HGSS (set1 is worse than set4 often enough that I am comfortable in just stating this as a fact).
But Platinum Lvl100 / open level is _slightly_ harder than HGSS.

Thorton is not that deadly though, and not a significant step up from other trainers with smart AI, particularly the tier 7 trainers poking around in 42-48.

Lvl 50 remains the easiest on both games; this is because of set predictability + the fact that tier 1 + tier 2 pokemon are so bad that the AI almost cannot win with them (compare this to open level's round 1 0IV garchomp1 where the AI can totally luck into clicking EQ twice and winning).
That said, the actual difficulty difference is very small for long streaks. This is mostly a nod to how much easier it is to get to around 35-40 on lvl50 compared to open level.
So what next?
I don't know. It's possible I change my mind on a lot of the above, perhaps all. This serves as a decent snapshot of where my mind is at on factory for now.

While I'm not looking for an active discussion on much of the above (because in many things I think my mind was kind of already made up a very long time ago), I encourage people to challenge the above and think critically on what parts you do/ do not agree. The two failures of this guide would be if it was ignored completely and if it was blindly believed word-for-word.
 
Last edited:
Ahead of my factory doubles writeup I do want to dump my thoughts into a rough guide continuation.

1. "Dumb AI" (rounds 1-4) in doubles is significantly smarter and while its targetting choice can be random, its move choice is a lot stronger. This makes reaching 21 or even 28 rather difficult. Equally, its a waste of time to take ages on battles 1-14. So get ready to lose a lot.
2. While the very very best doubles drafts are better than the best singles drafts, things can go wrong in doubles a lot easier and on average your doubles draft is incredibly likely to have significant holes in it.
3. Thorton doesnt appear in doubles, but dont think this makes doubles any easier. I would say doubles is harder although I'm probably a weaker doubles player.
4. I think Lvl50 doubles is easier at the start than Open Level, but harder after 49+. The small increase in raw damage output tips the AI towards making more "smart" decisions to go for guaranteed KOs. (In singles i think lvl50 is easier throughout).
5. I do NOT recommend going for a lot of swaps just for elevations in doubles (both lvl50 and open level but moreso in open level). The early AI is too strong to just recklessly swap spam like you can do in singles. If the decision is close then you can swap, but do not swap knowing the draft is getting weaker.
1. Speed is the most important stat by a massive margin. This is obvious/well known but in the factory you can almost multiply how important it is by another 50% or something. A singles draft can have room for slow pokemon and gen 3 doubles can too due to the IV gap. In gen4 doubles, a slow pokemon is never that good and the acceptable ones are exclusively those that either function very effectively in a final 1v1 (as a 3rd pick) or those that are effectively immune to OHKOs (think; bronzong, etc).
2. Pokemon that are not OHKO'd by many moves get a huge leg up in doubles. This is because the AI's smartest protocol is to go for guaranteed KOs and the second it can't easily spot that, it starts to behave semi-randomly. This is a really good thing, because setup moves and a lot of status moves are quite often a waste of time from the opponent. So it's ok to be frail-ish but its not ok to drop too often. This strongly includes 4x weaknesses on slower sets.
3. Pokemon weak to spread moves (notably, fire/rock but also ground and electric among some others) are significantly better because the AI is incentivised not to hurt its teammate unless it sees a certain kill. A good example of this is Arcanine (even without intimidate) as it can survive a lot of Surfs and Earthquakes compared to singles.
4. As the inverse of 3, pokemon with spread moves are typically a lot worse. It's still usable but unless both of your leads are fast, intentionally hurting the teammate for non-trivial damage introduces quite a lot of variance in what could happen.
5. for 28+, you get quite a lot of information on turn 0 based on what you see. You can usually eliminate quite a lot of possible 3rd opponents - although equally it would be rare to not just make decisions based on whats immediately in front of you.
6. I would take "type" hints a lot more seriously in doubles. In singles, I generally advise against overreacting to most type hints unless your draft is terrible.
7. You really just want to be aware of as many interactions as possible. A lot more "weird stuff" happens in doubles is my experience. I guess "just have an encyclopedic knowledge of the game" sounds dumb but its kind of true. I guess the good news is that the best way to learn/improve is probably by playing factory.
8. "knowing every set" also helps a lot in cutting down the amount of realistic analysis you have to do. If you don't know every set, at least have a rough idea of how many sets run variable physical or special bulk - because then if you check that pinsir1 faints to flamethrower, you dont need to check for pinsir2/3/4. this is a significant timesave because you're probably doing some kind of calculation check for like 12+ moves, minimum (compared to singles where it maybe be like 3) in some scenarios).
1. Unless you get a blessed draft that allows it, I recommend against trying to build into any kind of "intentional" composition (even something as simple as EQ + flying or 2 slow pokemon + TR or "sun team", etc). This suffers from a similar effect to the "type synergy" team building attempts in singles, where you often make the composition worse by trying to force it, and you would have been better off to just take 3 good sets.
2. As mentioned in #3/4 above, I would shift pokemon weak to spread moves up a "tier" in whatever your rough mental estimation is and pokemon with spread moves down a tier.
3. I think its basically unacceptable to run 2 slow pokemon in the front, where we'll define slow as "70 base speed or lower without investment". Ideally you have at least 1 genuinely fast pokemon. Being slower than both opponents on turn 1 is at minimum a headache and regurlarly a loss.
4. Fast pokemon that are in some way hamgstrung by their set are a LOT better in doubles. Examples; pokemon with only 1 real attack (STAB overheat, leaf storm) are a lot better in doubles. So are pokemon with shaky accuracy (Garchomp2, Lucario2) because although you may need to click an 80% move, you at least can evaluate the option of double-targetting with a higher accuracy move.
5. Ideally there is some level of type synergy specifically on the 3rd pokemon, but I would not stress about this. To continue the Lucario example; a fire resist, fighting resist or a ground resist/immune would be helpful but not necessary as either double targetting or sacking is fine.
6. There is some level of compromise on the 3rd pokemon between being a really strong singles pick and between simply being fast. This remains a very difficult judgement call and I guess my conclusion would be to break it down on an individual level and try to properly assess the likelihood of either being ahead or equal on average given the leads.
7. In better drafts, be sensible/ cognizant of how the factory sets are split out on a speed basis (this is the doubles equivalent of the singles tip to never be weak to water). The deadliest things you can run into are always going to be the fastest, and the simple fact is that fire, electric, flying and to a lesser extent dragon/psychic dominate the faster sets. This is, in essence, the most important additional analysis step you'll do on a strong draft, which could be to drop a pretty good flying type simply because it will drop to too many faster thunderbolts. Equally its yet another reason to not draft a grass or bug type, and to stay away from fighting types that arent exceptionally strong in some other way.
8. The ultimate doubles move (protect) is rare in the frontier and too often is on already bulky sets. But the honorable mentions of fake out, endure and even fly(!) deserve a nod to the way that you can essentially cheat the ai on the first turn.
9. To further from #1, I would generally call spread damage teammate immunity a bait, not because its bad but because of how many pokemon with earthquake, surf, etc are just slow and bad. Although it is important to underline that "75" base power EQ is a much worse move, and you often dont massively care about doing double chip compared to just guaranteed taking something out.
10. I would not concern yourself with "all physical" or "mixed" or "all special" teams that much. It is very difficult to wall in doubles; 3 special attackers usually out-pressure something like blissey. You should be more concerned about limited type coverage across the 3 pokemon.
11. Outside of the above I think that the singles tier list holds up reasonably well. Similar to the spread damage point, i'd also maybe move some very fast pokemon up a tier or two, and move some of the slowest sets down even further.
1. The simplest tip I have is to take advantage of safe redirection. For example; there is a 5% HP golem and a 100% HP umbreon. You have a staraptor and a rapidash. In this situation and by default, you would target golem with both attacks even if your actual intention is for the 2nd attack to hit umbreon. This is because KOing golem is by far the highest priority and likely just a straight up wincon - and you don't want to lose because of a QC activation killing the pokemon intended to take it out, or because golem randomly survives via focus band (or alternatively, "something else" avoids death via lax incense/ brightpowder).
2. Doubles gets really complicated, especially once the enemy sets become "any" instead of fixed. My intuition on the play I want to make within 1-5 seconds of thinking is correct something like 70-80% of the time. But that other 20-30% in noticing alternatives and assessing sacking options, what the 3rd can/cant do, whether or not to err of the side of saving a lead, or whether to go for a double target KO on turn 2 vs priority takedown on turn 1 is hard. In my opinion "optimal" doubles play probably involves taking 10-30 minutes in non-obvious situations. I dont know anyone who plays this slowly in gen4, but it's probably the best single piece of advice to someone maybe breaking 100 eventually.
3. as mentioned in #4 for drafting, lower % plays are much more often correct in doubles and you have to accept this when playing. Hypnosis, Stone Edge or even straight up risking 50% chance for a pokemon to have inner focus on a fake out could all be totally valid and statistically correct decisions.
4. Another well known doubles point, but its usually best to take out the faster opponent if possible and if "all else is equal" (obviously do not leave something like golem or rhydon up if they would OHKO one of your sets towards the end of the turn and you have the option to shut them down first). The simple reason for this is that in an "even" match, you want the final 2v2 (or equiv.) to leave as many options open as possible, and you gain a lot of win% by simply being a bit faster than whatever the unknown 3rd set is. Equivalently, it may be obvious that the opposite (facing a 2v2 where you are slower than both opponents) is perhaps the worst possible endgame situation unless the win is trivial for some reason.
5. Also probably obvious to veterans, but be aware of useless opponents. Something like Shuckle can quite often be completely ignored and you can treat the game as a 2v1 into another 2v1 (and then kill shuckle). Weaker versions of this also hold true though, and the generalised advice is "Dont always OHKO something just because you can". Leaving Vileplume up for a turn is often fine, and the STAB psychic may be better spent on the other opponent.
6. Likewise in a matchup against 2 relatively unthreatening opponents, ideally you want to KO both on the same turn, "just in case" the last pokemon is a significant threat to win a 1v3.
7. Be very wary of explosion draws. The AI will happily click this move even if its guaranteed to faint its partner and cause a draw. This can make certain opponents a much higher kill priority than you may assume, although equally be aware that guaranteed regular KOs are still the higher priority.
8. Boosting moves very rarely have a chance to work in doubles. Even in scenarios facing 2 special attackers with a faster pokemon that runs calm mind, I think i'd often err against using it out of fear of a crit/ 2ry effect.
9. Conversely a lot of status moves - and particularly those affecting speed - are a lot better. Hitting a thunder wave, or preparing rain for a swift swim pokemon often beats everything except going for a guaranteed KO.
10. It's really important to be able to recognise sets by rough damage and implied speed. You may not have time to fully note down all of this, so make sure you keep a mental tab of roughly how much damage everything does, in case something gets double targeted and so the reverse-stat calc is harder to do. Although equally you can solve this problem by just recording/ streaming...
11. The AI's target decision (if it does not see a KO) is based on scoring but this often means its kinda random. It does not see "guaranteed" 2hko double-up KOs. It does not understand that dealing 70% to your Garchomp may be better than dealing 25% to your Lapras.
12. If you want to master the factory, then detailed decomp knowledge is required. https://github.com/pret/pokeplatinum/blob/main/asm/trainer_ai/trainer_ai_script.s it's relatively daunting to just learn or apply this constantly, but equally doing so probably would make you the best factory doubles player in the world so long as you are able to masterfully/strategically apply the insights. I have read it a few times but definitely do not "know" it and I am occassionally surprised or blunder based on it.
 
Battle Hall Doubles - 394 with Metagross (on emulator)
proof: 1-150, 151-300, 300-395

same teams as last time:
explode A (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 8 HP / 116 Atk / 172 SpD / 212 Spe
Adamant Nature
- Explosion
- Earthquake
- Rock Slide
- Thunder Punch

support A (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 160 HP / 72 Atk / 60 Def / 216 Spe
Impish Nature
- Gravity
- Rock Slide
- Protect
- Thunder Punch
explode B (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 40 HP / 252 Atk / 4 Def / 212 Spe
Adamant Nature
- Explosion
- Earthquake
- Rock Slide
- Iron Head

support B (Metagross) @ Focus Sash
Ability: Clear Body
EVs: 56 HP / 252 SpA / 200 Spe
Modest Nature
IVs: 2 Atk / 30 SpA / 30 Spe
- Hidden Power [Fire]
- Grass Knot
- Psychic
- Protect

Lost to ttar with team A.
Turn 1: I EQ+rock slide, putting both ttars in range to die from another EQ. The explosion metagross gets crit by crunch from one ttar, and the other ttar flinches. I should still win as long as quick claw doesn't activate for either ttar next turn.
Turn 2: Quick claw activates, and the explosion metagross dies to crunch. The support metagross lands rock slide on both ttars, putting them in the red. The other ttar uses crunch, putting the support metagross in range to die from another attack. I can still win if quick claw doesn't activate again and rock slide hits both ttars.
Turn 3: Quick claw activates again, and metagross dies to dark pulse

I've been thinking about replacing rock slide on the team A explosion metagross with iron head (or possibly brick break) to help better deal with situations like this.
Likewise, I think running thunder punch over rock slide on the team B explosion metagross helps cover more mons.

Gonna keep doing attempts to try and get the record.
 
Hi everyone!
First time posting in a forum since 2010, so let me know if I did something wrong and sorry if I can't write correctly to make the cool fancy text and pics like you do. I always loved the gen IV Battle Frontier but never experienced its gameplay to the fullest. When I played it as a kid my best knowledge of "serious" battling was understanding EVs and natures so I didn't go past the Silver Prints. A couple months ago tho I had a massive blast from the past and picked up my old Soul Silver to put an end to some challenges I left around for almost 15 years. And with my new knowledge and skills I finally got five stars on my trainer card and got four out of the five Gold Prints (fuck the Battle Factory, that shit is fucking outrageous). So here I am, submitting a couple records I did on my Italian retail Pokémon Soul Silver; I know these are rookie numbers and similar winstreaks are a child's play for you, but I don't care and I'm very proud of it anyways. Teams and brief analysis down below (IVs are 31 except where stated), with pics and battle videos as proof only of my highest winstreaks cause I'm stupid and thought of only keeping the best ones.

Battle Tower Singles winstreak: 148 (Gyarados, Jolteon, Bronzong), 121 (Gyarados, Jolteon, Scizor)

Gyarados LV.50 @ Lum Berry
Ability: Intimidate
EVs: 252 Atk / 4 Def / 252 Spe
Jolly Nature
- Waterfall
- Earthquake
- Ice Fang
- Dragon Dance

Jolteon LV.50 @ Focus Sash
Ability: Volt Absorb
EVs: 4 HP / 252 SpA / 252 Spe IVs: 14 Atk, 30 SpA
Modest Nature
- Thunderbolt
- Shadow Ball
- Signal Beam
- Hidden Power (Grass)

Bronzong LV.50 @ Leftovers
Ability: Levitate
EVs: 252 HP / 128 Def / 128 SpD
Brave Nature
- Gyro Ball
- Earthquake
- Toxin
- Protect

Gyarados is your standard booster that sets up with DDance and sweeps with the other three coverage moves. He also turns down opponents with Intimidate making much harder to OHKO him with physical moves. Lum Berry is huge cause para, sleep, freeze and confusion are extremely common and mean certain death in later rounds. Jolteon is just a glass cannon with a full damage/coverage moveset (Signal Beam is kind of a filler, almost never used it). The sash allows him to survive one hit and steals wins very often. I've never really considered Bronzong as a competitive Pokémon and oh boy if I was wrong. The guy is fucking amazing. I went for a full defensive tank set cause I needed something that could slowly but steadily kill very bulky mons like Cradily, Blissey, Snorlax, Walrein, Wailord, Miltank etc. I built him as a wall expecting to only do relevant damage with Toxic and to my biggest surprise, even without offensive EV investment Bronzie packs a real punch. 225 BP Gyro Ball hits really hard and Earthquake covers Steel and Fire. Basically the only two things that he can't really hurt are Skarmory and another Levitate Bronzong like him. Absolutely fantastic mon, I regret having slept on him for so long. This team's typing synergy is fantastic and swapping is core: Jolteon absorbs Electric moves that target Gyarados (while also refilling his own health re-enabling the Sash) and vice versa Gyarados is immune to Ground moves. Same with Bronzong that takes Rock moves aimed at Gyarados, and his only weakness to Fire is well tanked by Gyarados. This can turn off entire opponents and make most physical attackers a joke, like Electivire 4 that is supposed to be very threatening with a set that covers every memebr of the team. Just juggle with pokeballs until the opponent is at -6 Atk. In other cases, you can even simply PP stall your opponents.

Lost on battle 149 vs Walrein 3, Articuno 3/4, Glalie 3
Video: 14-81672-53610
I knew I'd lose to either hax or OHKO moves, since this team has basically no protection against them. Got haxed three times, each one differently took down one of my mons. Walrein's Sheer Cold kills Bronzong on turn 1, Articuno condemns Jolteon with an Ice Beam freeze, then Glalie lands a decisive critical hit on Gyarados. I should have swapped into Jolteon on turn 1 and tank the Sheer Cold with sash, maybe it would have ended differently but I just didn't think about it and simply misplayed. I'd definitely try Skarmory/Forretress with Vigor if I ever attempt this again, but I've been haxed enough in the past three months and I'm done for now. I really hope gen V Battle Subway won't be so mercilessly RNG dependant when I'll tackle it in future.

27553ad1-2392-43a4-9045-2126dab8383f.jpg

For my 121 winstreak (and Battle Castle/Arcade Gold Prints as well), same exact team but with Scizor instead of Bronzong.

Scizor LV.50 @ Occa Berry
Ability: Technician
EVs: 252 HP / 98 Atk / 160 SpD
Adamant Nature
- Bug Bite
- Bullet Punch
- Superpower
- Swords Dance

Scizor is a great mon and acts as a bulky setup sweeper. Sword Dance makes him hit really hard and land OHKO freely even with a low Atk investment. Technician Bullet Punch is a 90 BP with priority and Superpower provides massive damage and great coverage being a Fighting type. I eventually switched to Bronzong that, although packing lower direct damage, is much more durable and consistent for his defensive stats, better typing synergy and being able to heal with Leftovers.
I don't remember the others cause I easily KO'd them with Gayrados, but lost to a traumatic Hippowdon 4 that with Quick Claw outspeeds 2DD Gyarados before he could use Waterfall and kills him with a Thunder Fang crit (Hippowdon wasn't Intimidated so the crit was irrelevant, but still). Then Stone Edge crit on Scizor while he uses Swords Dance and RIGHT AFTER THAT another QC + Stone Edge crit outspeeds and kills Scizor. Finally the motherfucker, still at full health, tanks HP Grass remaining at about 25% health and hits Jolteon with Earthquake; the sandstorm damage nullifies his sash. Yep, lost in a 3v1, three crits in a row with 2 QCs. In later rounds of Battle Frontier facilities the AI is ridiculously unbalanced and some battles become a massive RNG nightmare. My soul was demolished after this one and I don't even know where I found the strength to start the Battle Tower again and go further.

(CONTINUES...)
 
Last edited:
Battle Tower Doubles winstreak: 281 (Bronzong, Swinub, Camerupt, Slowbro), 176 (Bronzong, Swinub, Camerupt, Octillery)

Bronzong LV.50 @ Occa Berry
Ability: Levitate
EVs: 252 HP / 128 Def / 128 SpD
IVs: 0 Spe
Brave Nature
- Gyro Ball
- Earthquake
- Trick Room
- Explosion

Swinub LV.1 @ Focus Sash
Ability: Oblivious
EVs: Any
IVs: Any
Any Nature
- Endeavor
- Protect
- Ice Shard
- Light Screen/Reflect/Odor Sleuth/Toxin/Sandstorm

Camerupt LV.50 @ Life Orb
Ability: Rock Solid
EVs: 4 HP / 252 Atk / 252 Spa
IVs: 30 HP, 30 Atk, 3 Spe
Quiet Nature
- Eruption
- Earthquake/Earth Power/Rock Slide
- Hidden Power (Grass)
- Protect

Slowbro LV.50 @ Sitrus Berry
Ability: Own Tempo
EVs: 252 HP / 252 SpA / 4 Spd
IVs: 0 Spe
Modest Nature
- Surf
- Psychic
- Ice Beam
- Trick Room

Bronzong + LV.1 is by far the strongest and most used opening in Doubles, everyone knows so yeah. Togekiss has the Follow Me advantage to redirect OHKO and threatening moves but my Bronzong has never been targeted by an OHKO on turn 1 in 500+ battles. Is it luck? Who knows. I prefer Swinub much more than others cause being Ice/Ground takes no weather damage, and I've found Abomasnow, Hippowdon and Tyranitar to be pretty frequent. In those cases you lose a lot of potential with Togekiss since you only take down one mon instead of two, and their weather lasts indefinitely. For Swinub Protect and Endeavor are almost the only relevant moves, overall in my battles I used Ice Shard twice and Light Screen once. You can even slap Fissure there it won't matter.
Bronzong is the same from my Singles team, except with Occa Berry, TR and Explosion. The former is core strategy, the latter is just so damn strong in general and even stronger in double battles. The defensive EV spread paired with Occa Berry allows Bronzie to only die to very strong Fire STAB crits ( generally for my experience it SURVIVES crit Fire Fang, Fire Punch and Flamethrower; DOES NOT SURVIVE crit Fire Blitz and Overheat, except by Typlosion 3 or Exploud 4 cause they're weak).
Camerupt is just a solid cleaner that kills everything with Eruption. Other moves are used less frequently but still great for coverage. Those 3 IVs grant the lowest speed obtainable with a 70 BP Grass Hidden Power, that I picked to deal with Quagsire, Swampert, Gastrodon. In my first winstreak of 176 I used a "tanky" Camerupt with full HP/SpA investment and Earth Power but in the end even with Rock Solid he couldn't stand any of the many super effective moves he's weak to, so I switched to the current mixed offense with Earthquake.
Slowbro is the backup TR setter and was the less frequently used cause most of the times battles end with Camerupt. Hits hard and with good typing moves. Surf if nice in doubles and since battles last a few turns, Sitrus Berry instead of Leftovers.

Lost on battle 282 vs Slowking 2, Mr. Mime 3, Bronzong 1, Slowbro 4
Video: 39-33123-73437
The first Quick Claw that killed Swinub before he could even move was damn crucial but other than that I lost because it was almost a mirror match with me in heavy disadvantage. I didn't even get haxed too much, but the opponent's team seems to be built *ad hoc* to face mine. I mean Bronzong, Slowbro and Slowking together? And two Trick Room users? Really? All that repeated setting and restoring the trick dimension was driving me crazy, and if only the opponent Bronzong had Fireproof instead of Levitate I would have won. It's a pity they managed to stop me this early, I'm really confident on this team and I'm sure it could have got a much higher winstreak. Also fuck Quick Claw, that shit's broken and I absolutely hate it.

0d1d1119-7f2e-4185-8c8d-e2cd891d165b.jpg

In my 176 winstreak I used Octillery instead of Slowbro, along with the same team but with the previously mentioned EV spread and moveset for Camerupt.

Octillery LV.50 @ Choice Specs
Ability: Sniper
EVs: 252 HP / 252 SpA / 4 Def
IVs: 0 Spe
Quiet Nature
- Water Spout
- Ice Beam
- Flamethrower
- Energy Ball

Not much to say about Octillery, he's really good in Trick Room and Choice Specs Water Spout can easily clean entire teams. Full attacker build for coverage but mostly filler since you almost always pick WS. Hits much harder than Slowbro with better coverage without even damaging allies but it's less bulky and most importantly can't set up a new Trick Room, and that's why I swapped him out. Otherwise he's like a stronger water Camerupt, fucking good.

Lost on battle 177 vs Moltres 4, Venusaur 3, Lapras 2, Whiscash 3
This was one of the only two in my 500+ double battles where my Bronzie didn't set the Trick Room cause he got instakilled by Moltres with a crit Overheat. Not having TR setup I know I'm already 90% fucked. In fact without TR they easily pulverize Swinub (Protect fails the coinflip). Camerupt deletes Venusaur with Eruption, gets killed by Whiscash's Hydro Pump. Octillery kills both Moltres and Whiscash with Water Spout. The last mon is Lapras and I'm locked on WS by the Choice Specs, I see that Lapras has Water Absorb and realize that it's over. He doesn't land any OHKO but time kills me anyway with Perish Song.

(CONTINUES...)
 
Last edited:
Battle Hall Singles winstreak: 246 (Raikou), 185 (Slaking)

Raikou ★ LV.100 @ Choice Lens
Ability: Pressure
EVs: 248 SpA / 36 SpD / 226 Spe
IVs: 30 HP, 5 Atk, 30 Def
Rash Nature
- Thunderbolt
- Shadow Ball
- Aura Sphere
- Hidden Power (Ice)

Raikou LV.100 @ Choice Lens
Ability: Pressure
EVs: 252 SpA / 4 Def / 252 Spe
IVs: 30 HP, 17 Atk, 30 SpA, 30 SpD
Modest Nature
- Thunderbolt
- Shadow Ball
- Extrasensory
- Hidden Power (Ground)

Not much to say about these guy since Raikou is the most used mon in the Hall, there are hundreds of posts about him and everybody has already extensively talked about its power and its threats. Thunderbolt and Shadow Ball, common in both sets, together cover Water, Flying, Psychic and Ghost. Aura Sphere kills Normal, Dark, Rock, Steel and Ice. HP Ice covers Grass, Ground and Dragon. Extrasensory hits Poison and Fighting. Finally HP Ground covers Fire and Electric. These movesets cover everything except Bug types, (which are mostly very weak), and Water-Ground types like Whiscash, Quagsire and Swampert (the formers don't appear after Argenta and Swampert isn't that much more dangerous than other mons that can be 2HKO anyways like him). Before Argenta I picked Ground and Water first, then no particular order for types, I just swapped event Raikou with the story one just when fighting Poison, Fighting, Fire and Electric, then after Argenta swapping Raikou every round.

Lost on battle 247 vs Gengar.
Video: 68-64119-02744
After picking Poison type, Gengar outspeeds Raikou and puts it to sleep on turn 1 with Hypnosis. Fuck me, if my Raikou was Timid instead of Modest this wouldn't have happened. Not too mad about it cause the Hall is the most "hit or miss" of Frontier facilities and this is still a fantastic streak.

29dee51c-669c-44a7-9d9b-88720d545652.jpg

Slaking LV.100 @ Choice Band
Ability: Truant
EVs: 4 HP / 252 Atk / 252 Spe
Jolly Nature
- Giga Impact/Return
- Earthquake
- Fire Punch
- Shadow Claw

OOGA BOOGA BIG MONKI HIT HARD.
Slaking
needs to kill the opponent in one turn if possible to not fall in Truant and be doubly exposed to dangers. I chose Giga Impact over Retun cause the higer power does really make a difference and the recharge turn doesn't stack with Truant. STAB Giga Impact is a 225 BP that with Choice Band scales off ~630 Atk. Hits like a fucking truck. Stronger than super effective Earthquake, a neutral hit will OHKO any opponent, always with the exception of Sash/Band/misses etc. Return has 100% accuracy and usually can't miss (again with exceptions), but it can't OHKO bulkier pokes and with Truant that's a problem. I preferred a base 10% chance to miss than X% chance of not OHKO the opponent and in fact, paradoxically, GI with 90% acc was more consistent than 100% acc Return in killing the opponents; ultimately it's personal preference between these two moves. Earthquake kills Rock and Steel types that you can't stomp with GI, same with Fire Punch for Steel types like Skarmory, Scizor and Levitate Bronzong. Shadow Claw for Ghost types immune to GI (and together with FP, hits Shedinja). If I'm not wrong Shuckle and Spiritomb are the only mons in the Hall without Focus Sash/Band that you just can't OHKO, everything else goes down.
The run killer is Infernape that inevitably starts with Fake Out and then KOs with Close Combat. So, other than doing Fighting and Fire types first, no particular order to follow. Other threats are Hariyama (very dangerous with Fake Out + Vital Throw and Focus Band) and obviously any poke with Focus Sash/Band, Protect, Bright Powder and Lax Incense. At first I also feared two-turn moves with invulnerability like Dig, Dive and Fly but the AI doesn't seem to know they can lock Slaking to death and will not prioritize choosing them.
Lost on battle 186 vs Tentacruel that with Bright Powder dodged two Earthquakes in a row and finished me with a crit Poison Jab.

ca5464a8-2442-4476-b9d7-c513d74c9c29.jpg

And that's all, sorry again for any errors. It's not much, but it's honest work.
Good luck with your Battle Frontier records and have a nice one!
 
Last edited:
Facility: Battle Tower Link (human partner) on Pokemon Diamond Version and Pokemon Platinum Version with tradeback movesets

Length of your streak: 56 (ongoing)

Method of Play: Retail

Team (Diamond)

Salamence @ Lum Berry
Male
Ability: Intimidate
EVs: 252 Atk / 4 Def / 252 Spe (Probably, I raised this thing over a decade ago using Smogon guides)
IVs: Not accounted for
Jolly Nature
- Dragon Dance
- Outrage
- Earthquake
- Flamethrower

Empoleon @ Chople Berry
Male
Ability: Torrent
EVs: Not accounted for (It's the starter you get from Rowan and I don't remember if I EV trained him properly before hitting level 100)
IVs: Not accounted for
Bold Nature
- Stealth Rock
- Hydro Pump
- Ice Beam
- Protect

Team (Platinum)
Arcanine @ Choice Band
Female
Ability: Intimidate
EVs: 252 Atk / 4 SpD / 252 Spe
IVs: Not accounted for
Jolly Nature
-Flare Blitz
-Iron Head
-Thunder Fang
-ExtremeSpeed

Tyranitar @ Leftovers
Female
Ability: Sand Stream
EVs: Not accounted for (used to be a niche Fling setup but I changed it so I'm not sure of the EV spread)
IVs: Not accounted for
Impish Nature
-Crunch
-Earthquake
-Stealth Rock
-Protect

I employed a basic strategy of declawing anything physical with double Intimidates, then had Arcanine take out any threats to Salamence in order to use Dragon Dance at least once. If any opponents knew Ice moves, I'd switch into Empoleon and eliminate them with Arcanine. Anything with a OHKO move was instantly targeted without any Dragon Dance setup. Protect and Salamence's typing meant I could usually use Earthquake to take out multiple threats. Stealth Rock on two different Pokemon is probably unwise, but Hydro Pump alone is a big gamble and I have been let down by Stone Edge and Rock Slide far too often to trust them again.

I will admit RNG was a big factor carrying this team. Outrage can't reliably target the mon you need to take out, and Dragon Claw is probably too weak to get the same results. There was also something going on with the AI because I would often see major misplays, like Poison Fang on Empoleon and Dark Pulse on Tyranitar (I'm not talking about switch-ins either). I just got the streak so I'll save speculation for later (I also haven't read all the guides so I figure there's a good chance y'all cracked the code) but since there are fewer than 10 users on the leaderboard and people might start playing at a moment's notice, I figured I'd post my streak now and get my fifteen minutes of fame while I still can. I don't know if this counts as a Platinum streak since the Trainers I fought were exclusively from the Diamond/Pearl pool, but it's at least eligible for the DP Tower.

Edit: I wrote Multi instead of Link for the format when I originally posted it.
 

Attachments

  • Multi Battle Tower A.jpg
    Multi Battle Tower A.jpg
    350.9 KB · Views: 27
Last edited:
Hello everyone!
I'm going to be updating my Garchomp's Hall streak which i never really posted before but mentioned in my Raikou run (still the longest streak here almost a year later).

So this is now a 252 battles won in a row Garchomp streak in the Singles category of the Battle Hall. Once again done on my Spanish retail copy of Heart Gold.

The Garchomp in question is Jolly natured with a Focus Sash as the item. The IVs are all 31 except for Special Attack, with 252 IVs in Attack, 252 in Speed and 4 in HP. Its moves are the following: Outrage, Earthquake, Fire Fang and Sword Dance.

About the order in which types were tackled. As always you try to beat the tough type first so you waste the least possible amount of time.

-Before Gold Print (Second Battle with Argenta at the 170th battle): ICE/ DARK/ WATER/ GROUND/ GRASS/ BUG/ FLYING/ NORMAL/ FIGHTING/ PSYCHIC/ GHOST/ ROCK/ STEEL/ DRAGON/ FIRE/ POISON/ ELECTRIC.

-After Gold Print (More or less the reversed order, from the easier to the tougher types): FIRE/ ELECTRIC/ POISON/ DRAGON/ STEEL/ ROCK/ GHOST/ PSYCHIC/ FIGHTING/ NORMAL/ FLYING/ BUG/ GROUND/ GRASS/ WATER/ DARK/ ICE.

I don't have a lot more to add apart from what has been said in other Garchomp runs or even my previous posts. I fell like this time i had a lot more knowledge of the strategies used here (acquired during the Raikou runs) as before them when using Garchomp i couldn’t even go past the 180 battles (and that’s why i started using the electric tiger).

Now i knew to avoid salac' s activations such as against Heracross ( using a weak Earthquake first and then finishing with Outrage) or recklessly getting auto-confused after using Outrage (For example against Venusaur which tends to protect and thus the confusion appears before doing the necessary damage, so if i hit myself then between the poison from toxic and direct damage from frenzy plant it can, and has even killed me before. Not anymore after using Earthquake first).

There were also other times where Earthquake (EQ) was preferable if only because it lacks the contact that Fire Fang and Outrage provide when attacking, thus preventing the damage dealt from abilities such as Poison Point or Sharpedo's Rough Skin which would negate my Focus Sash.


This time Slowbro was the culprit that ended the run. It survived the Outrage and brought me to 1 HP with Ice Beam. Because he recovered some of its life with a Sitrus Berry it could perfectly survive the second Outrage and finish me with Psychic.

I think that only a Critical Hit on my favor could have made things different, so not really angry at all.
Oh!, and those of you who might have read my previous posts might have realize one thing in common about the way my luckiest runs tend to end in this facility. Yeah, that's right! from my 5 best streaks so far, 4 of them ended after 252 victories and every time against a Water Type Pokémon. 4 times using Raikou (against a Swampert at the 355th battle, and another Swampert, a Sharpedo and a Starmie at the 253rd on the other ocassions) and now 1 with Garchomp where i lost against a Slowbro.

After this "fun" fact about my curse, i must conclude by saying that in the end i'm glad that i could give this Garchomp a pretty decent run (the 2nd best from its species as of today) as without trying him first last year and seeing that the gold was attainable i wouldn't have ever tried to get the top spot. And even though last time i got burnt out using him and not even today i could beat Nickm65's run with him, this shark was crucial in giving me the knowledge and courage to try the Raikou that in the end got me that desired World Record. So i see this as Garchomp opening the path to Raikou's fated run and now, with the experience gained there, i can finally give him a better deserved place for such a powerful and fun Pokémon to use.
 

Attachments

  • 20240701_171820.jpg
    20240701_171820.jpg
    2.2 MB · Views: 34
  • 20240701_184332.jpg
    20240701_184332.jpg
    2.3 MB · Views: 39
Hello everyone!
I'm going to be updating my Garchomp's Hall streak which i never really posted before but mentioned in my Raikou run (still the longest streak here almost a year later).

So this is now a 252 battles won in a row Garchomp streak in the Singles category of the Battle Hall. Once again done on my Spanish retail copy of Heart Gold.

The Garchomp in question is Jolly natured with a Focus Sash as the item. The IVs are all 31 except for Special Attack, with 252 IVs in Attack, 252 in Speed and 4 in HP. Its moves are the following: Outrage, Earthquake, Fire Fang and Sword Dance.

About the order in which types were tackled. As always you try to beat the tough type first so you waste the least possible amount of time.

-Before Gold Print (Second Battle with Argenta at the 170th battle): ICE/ DARK/ WATER/ GROUND/ GRASS/ BUG/ FLYING/ NORMAL/ FIGHTING/ PSYCHIC/ GHOST/ ROCK/ STEEL/ DRAGON/ FIRE/ POISON/ ELECTRIC.

-After Gold Print (More or less the reversed order, from the easier to the tougher types): FIRE/ ELECTRIC/ POISON/ DRAGON/ STEEL/ ROCK/ GHOST/ PSYCHIC/ FIGHTING/ NORMAL/ FLYING/ BUG/ GROUND/ GRASS/ WATER/ DARK/ ICE.

I don't have a lot more to add apart from what has been said in other Garchomp runs or even my previous posts. I fell like this time i had a lot more knowledge of the strategies used here (acquired during the Raikou runs) as before them when using Garchomp i couldn’t even go past the 180 battles (and that’s why i started using the electric tiger).

Now i knew to avoid salac' s activations such as against Heracross ( using a weak Earthquake first and then finishing with Outrage) or recklessly getting auto-confused after using Outrage (For example against Venusaur which tends to protect and thus the confusion appears before doing the necessary damage, so if i hit myself then between the poison from toxic and direct damage from frenzy plant it can, and has even killed me before. Not anymore after using Earthquake first).

There were also other times where Earthquake (EQ) was preferable if only because it lacks the contact that Fire Fang and Outrage provide when attacking, thus preventing the damage dealt from abilities such as Poison Point or Sharpedo's Rough Skin which would negate my Focus Sash.


This time Slowbro was the culprit that ended the run. It survived the Outrage and brought me to 1 HP with Ice Beam. Because he recovered some of its life with a Sitrus Berry it could perfectly survive the second Outrage and finish me with Psychic.

I think that only a Critical Hit on my favor could have made things different, so not really angry at all.
Oh!, and those of you who might have read my previous posts might have realize one thing in common about the way my luckiest runs tend to end in this facility. Yeah, that's right! from my 5 best streaks so far, 4 of them ended after 252 victories and every time against a Water Type Pokémon. 4 times using Raikou (against a Swampert at the 355th battle, and another Swampert, a Sharpedo and a Starmie at the 253rd on the other ocassions) and now 1 with Garchomp where i lost against a Slowbro.

After this "fun" fact about my curse, i must conclude by saying that in the end i'm glad that i could give this Garchomp a pretty decent run (the 2nd best from its species as of today) as without trying him first last year and seeing that the gold was attainable i wouldn't have ever tried to get the top spot. And even though last time i got burnt out using him and not even today i could beat Nickm65's run with him, this shark was crucial in giving me the knowledge and courage to try the Raikou that in the end got me that desired World Record. So i see this as Garchomp opening the path to Raikou's fated run and now, with the experience gained there, i can finally give him a better deserved place for such a powerful and fun Pokémon to use.
Hi!
Once again here, with a Hall Singles streak. This time the protagonist is Latios. 178 battles in my (also Spanish) Platinum retail copy.

I might provide more context in a future update as i still have some unfinished streaks on other facilities that i may end up lumping together in a single post.
But since this facility holds more importance to me i'll give it the spotlight next to the previous one with Garchomp.

Let's start by saying this is the very same Latios that won me all the Gold Symbols in Emerald (including a World Record Streak on the Pyramid), so it was only natural to try the same thing onto the next Battle Frontier. And so we did (again, i'll explain more in the future) leaving this facility for last as i already had beating it with Garchomp and Raikou.

What motivated me to beat it a 3rd time was not only to provide this dragon with one last accolade, but also seeing the lack of Latios attempts here (after trying it i might understand why xD). So with some hesitation and the determination to be the first one to beat it i started preparing the different sets.

I was really straight forward with the process as first of all I changed Latios EV's from those used in gen 3 to a more standard 4 HP/ 252 SpA/ 252 Speed.

This latios is Modest and its IVs are 29 HP/21 Atk/ 15 Def /31 SpA /23 SpD /29 Spe.
The item I equiped him with was the Choice Specs.

After cloning him in Emerald i gave him 5 different movesets (which will be named from A to E) but in the end i consider them as the same Pokémon trained differently for better performance depending on the type. But my headcannon aside i recommend doing the same thing for simplicity and effective resource management. The sets are the following:

-Latios (A): Psychic, Thunderbolt, Dragon Pulse, Surf
(ICE/ ELECTRIC/ GROUND/ STEEL/ POISON/ FIRE/ FIGHTING)

-Latios (B): Psychic, Thunderbolt, Draco Meteor, Surf
(DARK/ NORMAL/ ROCK)

-Latios (C): Psychic, Thunderbolt, Dragon Pulse,
IceBeam.
(GRASS/ DRAGON/ FLYING/)

-Latios (D): Psychic, Thunderbolt, Dragon Pulse, Energy Ball
(WATER)

-Latios (E): Psychic, Thunderbolt, Dragon Pulse, Shadow Ball
(BUG/ GHOST/ PSYCHIC)

Before the 170th battle: ICE (A)/ DARK (B)/ STEEL (A)/ ELECTRIC (A)/ BUG (E)/ ROCK (B)/ WATER (D)/ GROUND (A)/ PSYCHIC (E)/ NORMAL (B)/ GRASS (C)/ GHOST (E)/ FIGHTING (A)/ FIRE (A)/ POISON (A)/ FLYING (C)/ DRAGON (C)

After the 170th battle: DRAGON/ POISON/ FIRE/ FIGHTING/ FLYING/ GROUND/ PSYCHIC/ NORMAL/
GRASS / BUG/ ROCK/ WATER/ GHOST/ ELECTRIC/ STEEL/ DARK/ ICE

One thing to note is that i prepared all 5 of them before using them so now, for example, i might see the D set as unnecesary as i only wanted the Energy Ball for the water/ground types that still gave me nightmares from using Raikou but thinking it through now, i see they are manageable with psychic so WATER could go with GROUND and the others onto the A set (reducing the total number to 4).

Also after battle 170 i really didn’t know what to use so i opted for set-B as it had the coverage that i considered standard while having the more powerful but riskier option with Draco Meteor on the dragon STAB.

So just to be clear, a lot more thought could be put here, from the EV optimization to other different movesets, the items or even the orders to tackle the types (as i never really did a full run of the chart after gold).

But in the meantime, until me or somebody else decides to better plan this, i'll leave this here as the stepping stone to be improved in the future. Feel free to try it and tell me how it goes.

RUN ENDERS:

-Weavile, of course.

-Tyranitar outspeeds and kills with crunch

-Magnezone barely survives the surf and kills
with mirror coat.

-Meganium?!! (VS Argenta, battle for silver): completely my fault as i quicly clicked the light blue button expecting it to be ice beam when i was using a surf set. It paralyzes me with body slam and kills with a crtical energy ball.

-Scizor survives TB and kills with x scissor.

-Cacturns ends the run with sucker punch (This was on the 179th battle, so the one who ended the longest streak yet).

CLOSE CALLS:

-Honhckrow outspeeds but Latios survives and kills with TB

-Kingler survives but is paralyzed.

-After gold: houndoom outspeeds and crunches at latios who barely survives before thunderbolting.

- Togekiss triess to paraflinch after surviving the first Thunderbolt but fails.


Final Streak: 178 battles

I'll end the post talking about the satisfaction of seeing Latios having accomplished this fantastic and rough journey, with its name being now next to the ones of its teammates who had previously done some fantastic runs. And although this one seems much more modest (how appropiate), it gave me the opportunity to not only relay on previous tactics (which i have done mostly until now) but to try my own ones here.

So that's what i'll encourage you to do now, to not be terribly afraid of trying new things even though they might seem far from perfect. If you think is good enough, have fun with it!

See you in future updates!
 

Attachments

  • 20240703_213508.jpg
    20240703_213508.jpg
    3.7 MB · Views: 21
  • 20240703_212449.jpg
    20240703_212449.jpg
    3.5 MB · Views: 21
  • 20240703_212649.jpg
    20240703_212649.jpg
    2.2 MB · Views: 23
Last edited:
Hall Doubles - Metagross 492 on emulator
(video proof: 1-210 211-390 391-493)

Wanted to make a brief post about my current Hall Doubles PB since atsync said they were updating the leaderboard soon.
The teams have changed slightly since my last post:
explode A (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 8 HP / 116 Atk / 172 SpD / 212 Spe
Adamant Nature
- Explosion
- Earthquake
- Iron Head
- Thunder Punch

support A (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 160 HP / 72 Atk / 60 Def / 216 Spe
Impish Nature
- Gravity
- Rock Slide
- Protect
- Thunder Punch

Team A now covers Flying, Ice, Electric, Ghost, Psychic, Water, and Fire types. explode A is EV'ed to survive eruption from entei and outspeed starmie/raikou. support A is EV'ed to survive crit explosion from its teammate and to outspeed its teammate. explode A now runs iron head instead of rock slide.
explode B (Metagross) @ Choice Scarf
Ability: Clear Body
EVs: 48 HP / 252 Atk / 208 Spe
Adamant Nature
- Explosion
- Earthquake
- Thunder Punch
- Iron Head

support B (Metagross) @ Focus Sash
Ability: Clear Body
EVs: 56 HP / 252 SpA / 200 Spe
Modest Nature
IVs: 2 Atk / 30 SpA / 30 Spe
- Hidden Power [Fire]
- Grass Knot
- Psychic
- Protect

Team B now covers Dark, Bug, Grass, Normal, Rock, Steel, and Ground types. explode B is EV'ed to outspeed aerodactyl. support B is EV'ed to hit 225 speed (outspeeding things like magneton, skarmory, politoed, etc.). explode B now runs thunder punch instead of rock slide.
Pre-170: WATER (A), FIGHTING (B), GHOST (A), GROUND (B), ROCK (B), FIRE (A), STEEL (B), BUG (B), DARK (B), GRASS (B), FLYING (A), ICE (A), NORMAL (B), PSYCHIC (A), POISON (A), ELECTRIC (A), DRAGON (A)

this probably isn't "optimal" but I don't think the order really matters all that much pre-170.

Post-170:
171-180DragonPoisonFightingDarkBugGrassNormalRockSteelGroundB
181-190FlyingIceElectricGhostPsychicFireWaterFlyingIceElectricA
191-200DragonPoisonFightingDarkBugGrassNormalRockSteelGroundB
201-210GhostPsychicFireWaterFlyingIceElectricGhostPsychicFireA
211-220DragonPoisonFightingDarkBugGrassNormalRockSteelGroundB
221-230WaterDragonPoisonFlyingIceElectricGhostPsychicFireWaterA
231-240FightingDarkBugGrassNormalRockSteelGroundDarkBugB
241-250DragonPoisonFightingFlyingIceElectricGhostPsychicFireWaterA
251-260GrassNormalRockSteelGroundDarkBugGrassNormalRockB
261-270DragonPoisonFightingFlyingIceElectricGhostPsychicFireWaterA
271-280SteelGroundDragonDarkBugGrassNormalRockSteelGroundB
281-290PoisonFightingFlyingIceElectricGhostPsychicFireWaterFlyingA
291-300DragonPoisonFightingDarkBugGrassNormalRockSteelGroundB
301-310IceElectricGhostPsychicFireWaterFlyingIceElectricGhostA
311-320DragonPoisonFightingDarkBugGrassNormalRockSteelGroundB
321-330PsychicFireWaterFlyingIceElectricGhostPsychicFireWaterA
331-340DragonPoisonFightingDarkBugGrassNormalRockSteelGroundB
Repeat at 341

Ultimately the streak lost to an omastar with team A. In hindsight, I think team A is very poorly equipped to deal with omastar (although I still think I misplayed in the moment). I tried to double thunder punch one of the omastars, which would've needed very high rolls from both metagross's to possibly KO. I'm tempted to try and see if I can fit grass knot onto one of the team A metagrosses for future runs.

~~~
In case anyone is interested, here are some of my other attempts between my last post and now:
1: 1-170 171-310 311-449
2: 1-170 171-320 321-390
3: 1-170 171-280
4: 1-200 201-329
5: 1-170 171-339
6: 1-230
7: 1-200 201-229
~~~
I intend to keep grinding this until I beat Jumpman's record of 620.
 
Battle Hall Research: Opponent appearance rates
credit and thanks to the following users for their part in discussing, assisting, providing information: atsync Squilliams dankHug average fella


For a long time, users have noticed the opponents that you face in the Hall are definitely not uniformly random (this would mean that if there are 10 pokemon possible under type A at rank 10, then each have a 10% chance of appearing). Some pokemon are suspiciously very common, and some seem like they almost never appear. This is obviously something that we care about a lot for streaks. For example; Garchomp obviously cares a lot about how common Abomasnow or Weavile are. If they were a lot rarer (or a lot more common) this would be a massive buff or nerf to Garchomp's expected winrate past 170.

For those with a more formal mathematical background, I'd say that the data brought up by atsync in particular in the discord pretty clearly immediately disproved the null hypothesis (uniform-random). It also seemed to contradict a lot of other "reasonable" guesses (example; that the enemy trainer is random, but the pokemon that they can randomly generate are not).

To be honest; i struggled to notice any kind of coherent pattern. I looked at the held items of common/uncommon pokemon. I looked at if there was any connection in BST, in stats, in primary/ secondary typing, etc.
That said, if anyone has a really good idea that explains the data I'm about to bring up; please say so!


In the absence of any actual idea it was obviously going to be necessary to gather a lot of data to be reasonably sure about conclusions. Certainly too much effort to expect from humans, and besides this kind of manual tracking is immensely boring. But the battle hall does offer a uniquely simple setup: no switching, limited strategy, and a relatively easy UI, plus it can all be done via the touchscreen.

So the solution I came up with was simple too: just use a bot to play a ridiculous number of games in the hall, following how a human would play as closely as possible while trying to avoid any obvious biases. There were some potential alternatives to this, such as savestate loading/ trying to manipulate the rng in memory/ etc. But I rejected these either because I was concerned they would interfere with the results or because (frankly) they seemed higher effort or would take longer. The actual hall bot was ready in hours.

The Hall Bot (to make it abundantly clear, this is not a submission!!! and players should never cheat in a real submission):

Stage 1: Raikou

The first test subject was Raikou@specs with the same moveset that Carlos used for the current WR, although I made a few small changes to the EVs that honestly don't matter.
Raikou was chosen for a few reasons; first it seemed clear to me that with specs, the strategy should be "simple" (meaning no disrespect to any raikou players). You select 1 move, there is a "correct" move for each pokemon, and tbh its kind of easy to work out what the optimal move to pick is, even factoring in things like using tbolt (instead of the super effective move) against opponents with sash, for the increased win chance through paralysis.
The bot played 1,100 games with Raikou and its first actual streak that went past 170 hit 342 (!) I know there has been a small amount of grumbling in the past about the legitimacy of a few hall records; I think a bot hitting 342 with Raikou pretty much validates all Raikou records anywhere in this range (keep in mind; this raikou was not leaving water/ground to last like a human would). Raikou lost to an earth power crit from camerupt.
However some issues with Raikou arose:
- the winrate isnt really good enough. I mostly care about rank 10, and Raikou spends too often going back to 1-170.
- The bot had some slight errors with reading opponent names. Some of this was my fault, some wasnt.
- The bot was deterministically going through types in a set order. This is very undesirable because it means if e.g. we always face ground before steel, then the chance of seeing Steelix in steel are lower than it "should" be (since it could turn up in ground first and block itself).

The Raikou data was flawed (this become quickly obvious when i started reviewing the footage) and unfortunately I didnt know how flawed it was exactly; so its not a part of this report going forward.

Stage 2: Jolteon
I wanted a set that would not lose but also did not want to play on any kind of altered rom out of fear of in some way affecting the results. The solution was a bold Jolteon, with maxed (illegal) EVs, no guard, focus sash, sheer cold/ focus blast/ zap cannon/ (sacred fire/ blizzard). This thing is obviously deeply unfair. In fact pretty much the only way it can possibly lose is either the bot misreading the name of the opponent, the bot accidentally hitting "quit" (happened once), or exactly scyther/ scizor getting a swarm crit into a priority move (although they still have to dodge the status win% that jolteon gets through burn/paralysis). In hindsight, a superior version of this is probably replacing Jolteon with Raikou, but that doesn't really matter. This set is strong enough to not care about what type its clicking and can go through types randomly.
Jolteon went something like 7200-2 and collected a ton of data.
Things were not perfect however:
- Jolteon ruins the data collecting for electric types, as it prevents itself from appearing. To get accurate data on electric types, we need to use a non-electric
- Squilliams shared some of his data from doubles Hall with me and it occurred to me that maybe doubles hall operates differently to singles hall.

If you want to watch the bot play Jolteon, here is a sample:

Stage 3: Slaking Doubles
With electrics banned, its surprisingly hard to get a perfect alternative. Slaking came to mind instantly although i was not sure on no guard + OHKO vs. pure power. I decided that no guard dodges a significant amount of randomness (brightpowder, etc). Slaking can also run sacred fire to never lose to a weavile freeze, and Slaking's Special attack is also high enough to run zap cannon. The final move should probably be dynamic punch, since like focus blast it smacks the sturdy crew ridiculously hard.
So in summary: Slaking, Jolly, No Guard, Maxed EVs, Sheer Cold/ Dynamic Punch/ Sacred Fire/ Zap Cannon. The Slakings will be opposite genders and while one will hold a focus sash, i decided that the other holding lum probably overall increases its chances of not losing to random bad luck from something like QC dynamic punch Machamp or Hypnosis Gengar. It's not like anything kills Slaking apart from sash-counter (which wont work with the slaking double targetting in doubles...) or some obscenely powerful crit. While (illegal) Slaking is "worse" than (illegal) jolteon/raikou- the fact that it is doubles offsets this, as the opposing AI kind of sucks at doubles and also needs to get lucky "twice" to get through both Slaking.

If you want to watch the bot play Slaking:

Now it is definitely possible to improve the bot significantly. But this is about the hall, not any kind of hall bot. The bot only matters so long as it can serve its minimum function of reliably extracting data. I will at some point in the future return to this bot but have no immediate plans to do so for now.

The Results
As well as the bot results, I've included Squilliams' data into the spreadsheet since he linked it in discord.
Direct link for full transparency/ if anyone else wants to look into it: https://docs.google.com/spreadsheets/d/1-h21OWg2Vl4xylWgXVcIGU-oTEmnx-fmATbLq3cxEQI/edit?usp=sharing
I recommend everyone who cares makes a copy of this, you can add your own results to it if you like. The "ALL" spreadsheet breaks down by all types highest-> lowest and by each individual type.
Perhaps someone with more time or interest could try making this into a community effort. I'm not sure how "worth it" it would be though.

Percentage breakdown per-species, normalising for types (copying here just in case something happens to the spreadsheet linked above)
Altaria2.822309553
Steelix2.765068413
Banette2.354537473
Dusknoir2.226833746
Shiftry2.086206261
Dragonite1.905011298
Exeggutor1.83061319
Forretress1.78236075
Slowbro1.399361077
Yanmega1.361302081
Rotom1.342375884
Camerupt1.322327296
Skuntank1.302046124
Ninetales1.282256772
Roserade1.250942478
Umbreon1.227830832
Bastiodon1.212932615
Aerodactyl1.211453648
Scyther1.194858349
Lunatone1.158792648
Toxicroak1.150109594
Swalot1.147776184
Infernape1.100135966
Walrein1.089820391
Honchkrow1.073883124
Salamence1.03852605
Piloswine0.9937489982
Gallade0.990951715
Glalie0.9616925789
Lickilicky0.9494706974
Dewgong0.9290815879
Venusaur0.9156574078
Hitmontop0.9012633781
Magmar0.8975797403
Magneton0.8782111224
Torterra0.8682091291
Heracross0.8284919791
Probopass0.7896814718
Golem0.778111632
Rhyperior0.7701849935
Crobat0.7336913203
Bellossom0.7233704293
Noctowl0.7136303138
Kingdra0.7018189345
Venomoth0.6846037974
Aggron0.6657870347
Sharpedo0.6645101781
Ninjask0.6377639307
Sceptile0.6359300477
Snorlax0.6329804649
Ludicolo0.6286927841
Rhydon0.6254063086
Abomasnow0.6245803425
Luxray0.61514802
Manectric0.61514802
Armaldo0.5970563205
Scizor0.5716928008
Gastrodon0.5687730895
Grumpig0.5677587491
Gardevoir0.5677587491
Tropius0.5588148891
Pidgeot0.5539468752
Weavile0.5450356611
Omastar0.5444830476
Togekiss0.544412429
Blaziken0.5387539525
Raichu0.5382545175
Jolteon0.5382545175
Drifblim0.5349003356
Whiscash0.5199055358
Garchomp0.5192662121
Tyranitar0.5055774015
Heatran0.4985430509
Shuckle0.4974220515
Empoleon0.4968410197
Espeon0.4932985853
Flygon0.4867319871
Ursaring0.4801920768
Gengar0.4707885294
Charizard0.4637509843
Houndoom0.4570117076
Primeape0.4506316891
Poliwrath0.4503696431
Alakazam0.4467609829
Dodrio0.4332909053
Magmortar0.4247475557
Raikou0.4229142637
Cloyster0.4163419885
Sandslash0.4007052412
Vespiquen0.3982536711
Flareon0.3926911364
Staraptor0.38825807
Clefable0.3819709702
Donphan0.3766629268
Cacturne0.3748235034
Lumineon0.3591667332
Gliscor0.3523592828
Electabuzz0.3460207612
Meganium0.3418124006
Slaking0.338317145
Jynx0.3231510243
Articuno0.312354091
Swampert0.3121586258
Xatu0.3072493444
Hitmonlee0.305785789
Tentacruel0.3029502719
Cradily0.3024463353
Cherrim0.3020667727
Drapion0.295893377
Feraligatr0.2953148695
Weezing0.2949147139
Chansey0.2946633199
Zangoose0.2946633199
Magnezone0.2942935076
Vileplume0.2942038164
Crawdaunt0.2806150032
Bronzong0.2788732791
Lanturn0.2785694053
Huntail0.2713704206
Victreebel0.2705934055
Carnivine0.2702702703
Mr. Mime0.2699180938
Lucario0.2673531404
Tangrowth0.2623211447
Lapras0.2564187325
Metagross0.2506571051
Regirock0.2487761817
Golbat0.2475455541
Seviper0.2470907062
Glaceon0.2404231447
Ambipom0.2400960384
Slowking0.2314124901
Latias0.2312502648
Relicanth0.224613957
Mantine0.2237634774
Seadra0.2234815229
Zapdos0.2114571319
Entei0.2083667254
Nidoqueen0.2074545509
Girafarig0.2025378621
Kecleon0.1964422132
Claydol0.1928710987
Froslass0.1916208348
Starmie0.1875269628
Spiritomb0.183835439
Electivire0.1730103806
Moltres0.1682308689
Skarmory0.1631493703
Gyarados0.1597597921
Suicune0.1596296592
Stantler0.1527883881
Absol0.1444506861
Regice0.136239782
Blastoise0.1356852103
Octillery0.1356852103
Ampharos0.1345636294
Vigoroth0.1309614755
Persian0.1309614755
Tauros0.1309614755
Uxie0.1303052867
Breloom0.1285561735
Mamoswine0.1202115724
Porygon-Z0.1200480192
Dusclops0.1197222444
Rapidash0.1121974675
Jumpluff0.1118061796
Milotic0.1117407614
Arcanine0.1041833627
Fearow0.09960011673
Electrode0.09611687812
Wailord0.09577779551
Leafeon0.09538950715
Solrock0.0946878311
Registeel0.09278159213
Hypno0.08376768429
Hariyama0.08046994448
Nidoking0.07996730071
Kingler0.0798148296
Purugly0.07639419404
Cresselia0.07446016381
Floatzel0.07183334664
Porygon20.06548073775
Torkoal0.0641128386
Hippowdon0.0641128386
Vaporeon0.05587038072
Machamp0.04828196669
Rampardos0.04815022871
Golduck0.04788889776
Muk0.04782400765
Miltank0.04365382517
Granbull0.04365382517
Latios0.04252704478
Typhlosion0.04007052412
Politoed0.0399074148
Mismagius0.0399074148
Pinsir0.03979941097
Kangaskhan0.03274036887
Exploud0.03274036887
Kabutops0.031969487
Gorebyss0.03192593184
Mesprit0.02792256143
Azelf0.02792256143
Seaking0.02394444888
Lopunny0.02182691258
Blissey0.01091345629
Regigigas0.01091345629
Hitmonchan0
(Yes, I didn't see hitmonchan once in this sample of ~13,700 battles - but it did appear in atsync's data)

Warnings/ concerns/ criticisms:

I don't know /exactly/ what matters when it comes to opponent generation and RNG. It's possible that any of the following affected my data, in which case it may be different to yours:
- Trainer ID or other savefile-specific information
- System settings on the emulator/DS
- Species used in the battle hall
- Item used in the battle hall
- possibly the moves and stats matter although this would be surprising IMO.

I also haven't done any exhaustive analysis or quality control on the results, it's pretty much exactly what has been spat out of the bot. Potential problems:
- Pokemon with names very similar to others may have been misread. Porygon2 vs Porygon-Z for example.
- It's possible the bot has double-recorded a result
- It's plausible (although i'd bet against it) that the metronomic nature of the bot advances RNG in a much more predictable pattern than for a human player.
- It's possible I've made some mistake in the spreadsheet itself.

Despite this I will say that a lot of the data agrees (broadly) with Atsync's results in terms of pokemon that are common or uncommon under certain types.

~13.7k games is honestly not that many. Ideally we'd be collecting a lot of data from a lot of different players and combining it all. I don't ~really~ care enough to instigate that myself though, and I also think this kind of mindless data collection is best left to bots anyway.
I'm stopping at 13.7k because the bot takes a non-trivial amount of energy and time to run, and it feels like a lot of data. I guess ideally I could take it to 100k or beyond, but would that really matter? Do people really care if a pokemon is 1.1% to appear or 1.2%? If it was me i'd mostly be interested in approximate relative appearance rates.

What does this mean/ how do I use this?

Put simply, this potentially blows open the door for record breaking streaks.

Previously when preparing for e.g. normal type rank 10, you probably spent a lot of time looking at Blissey (indeed; blissey is a massive problem to specs Raikou!). But it seems that Blissey is almost a 0.2% appearance rate in normal. Of course 0.2% could happen, but to break the doubles record you "only" need 500 or so rank 10 wins after 170 - the overall chance that blissey turns up in those 500 games is something like 5%. Ideally you have a plan to beat blissey, but you kind of "don't need to".
Similarly ghost is almost entirely dominated by Banette + Dusknoir - and Rotom almost never shows up. But Rotom does show up a lot in electric. So if your pokemon struggles with rotom, it would only have to worry about it under electric.
Steelix meanwhile is a huge potential roadblock. It's the most common set under both ground and steel. There's no way you can ever consistently get past ~250 if you cant beat it. But on the flip side, anything which "always" beats steelix gets a massive leg-up from this; you should expect roughly 2-3 free wins per 100 games, that's a lot!

In fact looking at the results overall it does make a lot of sense why Raikou is sitting at #1 (even forgetting that the bot got 342 wins in a row with it) - the biggest "counters" to Raikou are just not common.

The top ~38 most common species make up over 50% of the overall appearance chance (this is higher than the other 169 sets put together).

It would still be a bold strategy to go into the hall with absolutely no answer to certain pokemon - but I'll call it "bold" rather than "stupid" because if you e.g. have no plan for Spiritomb, then you're actually likely to get away with it.

This raises another hall problem (and the original source of suspicion for long Garchomp/Raikou streaks which the above data largely proves are not cheated or at least are definitely reasonably possible); which is that more than any other facility you can simply "get lucky" in hall and not run into a counter. The most prophylactic thing players can do to avoid this also happening in the future is to simply record/ gather as much evidence as possible. Full videos are great, livestreams even better. It's an unfortunate reality of hall that the largest streaks emphasise luck a bit more. But on the other hand an "entire hall run" even at x1 speed should be hours at most unless the record is completely demolished.

For anyone aiming for a hall streak I'll once again bang the decomp drum; learning how the ai behaves is particularly important for hall (unless you want to do a load of direct testing in-game). There are several hall sets that look like they counter things, but the AI is completely incapable of using them properly (venusaur being a great example of this; just like i mention in the factory reports; the ai cannot "hyper beam" at 100% HP under normal conditions). This also applies to doubles; be aware of what the rules are for the use of moves like earthquake and surf. In certain scenarios these moves will never be chosen. Or, don't stop setbuilding just because something looks like it wins on first glance/ thought.
 
Hi!
Once again here, with a Hall Singles streak. This time the protagonist is Latios. 178 battles in my (also Spanish) Platinum retail copy.

I might provide more context in a future update as i still have some unfinished streaks on other facilities that i may end up lumping together in a single post.
But since this facility holds more importance to me i'll give it the spotlight next to the previous one with Garchomp.

Let's start by saying this is the very same Latios that won me all the Gold Symbols in Emerald (including a World Record Streak on the Pyramid), so it was only natural to try the same thing onto the next Battle Frontier. And so we did (again, i'll explain more in the future) leaving this facility for last as i already had beating it with Garchomp and Raikou.

What motivated me to beat it a 3rd time was not only to provide this dragon with one last accolade, but also seeing the lack of Latios attempts here (after trying it i might understand why xD). So with some hesitation and the determination to be the first one to beat it i started preparing the different sets.

I was really straight forward with the process as first of all I changed Latios EV's from those used in gen 3 to a more standard 4 HP/ 252 SpA/ 252 Speed.

This latios is Modest and its IVs are 29 HP/21 Atk/ 15 Def /31 SpA /23 SpD /29 Spe.
The item I equiped him with was the Choice Specs.

After cloning him in Emerald i gave him 5 different movesets (which will be named from A to E) but in the end i consider them as the same Pokémon trained differently for better performance depending on the type. But my headcannon aside i recommend doing the same thing for simplicity and effective resource management. The sets are the following:

-Latios (A): Psychic, Thunderbolt, Dragon Pulse, Surf
(ICE/ ELECTRIC/ GROUND/ STEEL/ POISON/ FIRE/ FIGHTING)

-Latios (B): Psychic, Thunderbolt, Draco Meteor, Surf
(DARK/ NORMAL/ ROCK)

-Latios (C): Psychic, Thunderbolt, Dragon Pulse,
IceBeam.
(GRASS/ DRAGON/ FLYING/)

-Latios (D): Psychic, Thunderbolt, Dragon Pulse, Energy Ball
(WATER)

-Latios (E): Psychic, Thunderbolt, Dragon Pulse, Shadow Ball
(BUG/ GHOST/ PSYCHIC)

Before the 170th battle: ICE (A)/ DARK (B)/ STEEL (A)/ ELECTRIC (A)/ BUG (E)/ ROCK (B)/ WATER (D)/ GROUND (A)/ PSYCHIC (E)/ NORMAL (B)/ GRASS (C)/ GHOST (E)/ FIGHTING (A)/ FIRE (A)/ POISON (A)/ FLYING (C)/ DRAGON (C)

After the 170th battle: DRAGON/ POISON/ FIRE/ FIGHTING/ FLYING/ GROUND/ PSYCHIC/ NORMAL/
GRASS / BUG/ ROCK/ WATER/ GHOST/ ELECTRIC/ STEEL/ DARK/ ICE

One thing to note is that i prepared all 5 of them before using them so now, for example, i might see the D set as unnecesary as i only wanted the Energy Ball for the water/ground types that still gave me nightmares from using Raikou but thinking it through now, i see they are manageable with psychic so WATER could go with GROUND and the others onto the A set (reducing the total number to 4).

Also after battle 170 i really didn’t know what to use so i opted for set-B as it had the coverage that i considered standard while having the more powerful but riskier option with Draco Meteor on the dragon STAB.

So just to be clear, a lot more thought could be put here, from the EV optimization to other different movesets, the items or even the orders to tackle the types (as i never really did a full run of the chart after gold).

But in the meantime, until me or somebody else decides to better plan this, i'll leave this here as the stepping stone to be improved in the future. Feel free to try it and tell me how it goes.

RUN ENDERS:

-Weavile, of course.

-Tyranitar outspeeds and kills with crunch

-Magnezone barely survives the surf and kills
with mirror coat.

-Meganium?!! (VS Argenta, battle for silver): completely my fault as i quicly clicked the light blue button expecting it to be ice beam when i was using a surf set. It paralyzes me with body slam and kills with a crtical energy ball.

-Scizor survives TB and kills with x scissor.

-Cacturns ends the run with sucker punch (This was on the 179th battle, so the one who ended the longest streak yet).

CLOSE CALLS:

-Honhckrow outspeeds but Latios survives and kills with TB

-Kingler survives but is paralyzed.

-After gold: houndoom outspeeds and crunches at latios who barely survives before thunderbolting.

- Togekiss triess to paraflinch after surviving the first Thunderbolt but fails.


Final Streak: 178 battles

I'll end the post talking about the satisfaction of seeing Latios having accomplished this fantastic and rough journey, with its name being now next to the ones of its teammates who had previously done some fantastic runs. And although this one seems much more modest (how appropiate), it gave me the opportunity to not only relay on previous tactics (which i have done mostly until now) but to try my own ones here.

So that's what i'll encourage you to do now, to not be terribly afraid of trying new things even though they might seem far from perfect. If you think is good enough, have fun with it!

See you in future updates!
And now my update on how the other facilites went.

For clarity I'll state the lenght of the streaks first, all of them were done on the Singles category and as always, on Spanish retail copies of Pearl, Platinum and Heartgold:

-DP (Pearl) Tradeback Battle Tower Singles: 145 wins.

- PtHGSS (HG) Battle Tower Singles: 102 wins.

-PtHGSS (Pt) Battle Castle: 171 wins.

-PtHGSS (Pt) Battle Arcade: 114 wins.

And now I'll present you the team that achieved those numbers, consisting of Latios, Garchomp and Raikou. These are their sets :

https://pokepast.es/40e402884ac990f2

Latios (M) @ Life Orb
Ability: Levitate
EVs: 4 HP / 252 SpA / 252 Spe
Modest Nature
IVs: 29 HP / 21 Atk / 15 Def / 23 SpD / 29 Spe
- Dragon Pulse
- Psychic
- Thunderbolt
- Surf

Garchomp (M) @ Focus Sash
Ability: Sand Veil
EVs: 4 HP / 252 Atk / 252 Spe
Jolly Nature
IVs: 1 SpA
- Outrage
- Earthquake
- Fire Fang
- Swords Dance

Raikou @ Choice Specs
Ability: Pressure
Shiny: Yes
EVs: 252 SpA / 4 SpD / 252 Spe
Rash Nature
- Thunderbolt
- Aura Sphere
- Shadow Ball
- Hidden Power [Ice]



As I haven’t seen anyone else using this exact combination yet, I'll take the liberty of nicknaming it. I'll refer to the assemble as "Syzygy" ( or "Sizigia" in Spanish) which is an astronomical term that describes the alignment of 3 or more celestial bodies. Which is exactlty how i felt when building it, as I started by wanting to see if it was possible to get the Gold Prints using my 2 already establisehd stars in their respective frontiers in conjuction at first, and then looking for a last member to complete the trio.

So Latios (the one that won me the Pyramid World Record in Hoenn) and Raikou (holding the current Hall World Record here in gen 4) were a lock in (probably my first "stupid" and mostly emotional decision), so I just had to find a third member that could help those 2 special sweepers together, and the first thing I did was to look at the other mon thad I had already bred and was battle ready, which was the Garchomp that i tried in my first, not so succesfull, Hall tries. It was mostly considered as he constituted a very powerful psychical threat to those walls that the other 2 would struggle with (mainly Blissey). He also felt appropiate as it could represent the Sinnoh side of the Battle Frontier, and that way i had a native species of each region where this challenge is present.

I know what you are thinking. Using 2 Dragons is quite risky and has already been utilized here (even the exact combination of Latios- Chomp) but usually with a better synergized third member, maybe a Steel type or something more bulky, and while i thought the same thing at first, the prospect of trying the limits of this 3 heavy-hitting monsters together sounded quite appealing.

I started by testing him again in the Hall and after having so much fun and getting a more decent streak (2nd best among Garchomp runs) he thus proved himself that he was able to shine along his soon to be teammates.

Now i had to truly try them together, and for that i didn’t have to really make any tweaks apart from giving Latios a more standard combination of EVs (4Hp/ 252 SpA/ 252 Spe), item(Life orb) and moveset (STAB Psychic and Dragon Pulse alongside Thunderbolt and Surf ) which were more appropiate for his role as a lead sweeper, similar to the one he had in Hoenn, although this time forgoing the option to set up with calm mind in favor of Surf if only to gain more coverage (specially for the ground types that would threaten Raikou).

Both Raikou and Garchomp were left as they were trained for the Hall, with their natures and Evs oriented to a similar sweeping role as Latios, but with their items being the factor in deciding their new duties now that they were part of a team.

Garchomp being the bulkier one and having the focush sash, lets him function as a mid to late game sweeper and (if the opportunity arose, or it was my only option left) I could also make use of my only set up option in Sword Dance and let him reap through whatever little team Latios left before dying.

Finally Raikou, which was much more limited in options due to its Choice Specs so it was more utilized as a sniper to clean up whatever resisted the slaughter of both dragons.

So the usual way that battles went was the following: Latios leads and takes out 1 or 2 members himself, leaving the rest for Garchomp's assault and if still necessary Raikou would clean the scraps.

Using this strategy we were capable of overwhelming almost every team by using the raw power of these beasts' sheer offensive capabilities.

The main threats for them were, of course, the Ice types (specially when they, very ofently, appeared in groups) leaded by Weavile, and some bulky mons, more strongly those capable of setting up or functioning under trick room (and thus negating one of my main advantages in speed superiority) with Bronzong being the most annoying one.

For the Ice types those two Thunderbolts were crucial for those who are also part water type such as Dewgong, Walrein, Lapras and Cloyster. The latter almost never presented a problem but the other 3 are bulky enough to survive Latios' first shock and getting their revenge later, usually needing to bring Raikou next, compromising my main strategy by locking Raikou in a move too early, potentially forcing me to break Chomp's sash if i wanted to circumvent it. Not to mention that they appeared together many times and that some of their sets are packed with OHKO moves. After the gold prints, Articuno presented the same problem.

This stacking of ice types against my own stacking of dragons was the main way most of my runs ended. Weavile was the most vicious one as he outspeed all 3 members and only Raikou's aura sphere was consistent enough to kill him when my tiger was at full health (and even then, I was at mercy of its higher crit chance with night slash and the possibility of freeze via ice punch) but if it were to be the set that holds focus sash i would also require the sacrifice of my Garchomp's very same item, thus meaning i depended on having both at full health.

In the cases where it was to appear as the lead i was almost certain to lose as 1 hp Garchomp would not only have to kill him through those freeze chances (which Raikou would have to avoid first along with the crit), but also take care of the 2 other mons who could very well be the aforementioned bulky water ones (who wouldn't die to Outrage unless it somehow critted and which would also need to hit 3 times or do the last one through confusion as it was the move needed to kill Weavile in the first place)

Another problem having to do with the secondary types of ice types, my available coverage and Raikou's choice specs.
Let's say that Latios faces one of this ice types as the lead (it usually was a glaceon or articuno that let me to these situations but could also be one of the bulky waters). Now i have to sacrifice him to chip it and leave that lead at the range of only one other attack for my shark or tiger to finish it and leave the battle as a 2 Vs. 2. I do this because I usually expected more ice types to be in the back (which was commonly the case specially if they were those Ace trainers in fur coats) but now i'm presented with a dilemma.

The sensible thing at first seemed to me to use Raikou, keep cleaning the field and leaving the sash user to finish the match (as it is also the faster of the 3). But if Weavile were to appear as the last mon, he is sashed and I face him with a crippled Raikou, the streak is finished. There is also the problem concerning which move i decide to lock my Raikou into. If it were to be TBolt that is usually stronger and sweeps the water/flying ones, but remember that Mamoswine exists and its inmmune to it. So the next option would be Aura Sphere, but wait! Frosslass is ghost, so another immunity roadblock. Also it doesn't even kill Abomasnow, whic also resists TBolt and its ability to set Hail on entrace breaks Garchomp's sash (Another big problem). So yeah the strategy is to better send Garchomp out first and then finish with Raikou if possible, but overall this whole ice fiasco is a terrible situation to be in (although that’s what i got myself into by insisting on using these 3 together).

Needless to say that most of these other mons were not a problem if Latios was still alive as Surf kills Mamoswine, Psychic most frosslass (except for one that outspeeds and i need the sash there), although Abomasnow survives and can kill/ sleep in revenge.

Now for Bronzong. I tried to kill him with Latios and with TBolt i could usually 3HKO him if he lets it happen but usually kills first if it doesn't get paralyzed. Usually because it sets the TR and that way it outspeeds and flinch with iron head. He also could Signal Beam and get very nasty confusions this way. Also this set can sleep via Hypnosis and kill with Dream Eater to the next ones.
If not Latios, the other 2 can also kill with TB (although this was the worst option as he could set the TR beforehand, kill latios and then kill Raikou with Earthquakes. Others could kill him with Explosion) or in the case of Garchomp with Earthquake or Fire Fang but yeahh, which one to use is up to luck because the well know combination of levitate and heatproof. If it was facing Garchomp i tried the fang first just to not completely waste a turn in case of ground immunity but if i sensed it doing very little damage i went all out with the EQ.

Other bulky Trick Room mons, mostly psychics and in combination with Bronzong itself sometimes could present a problem. Mainly claydol which could threaten Raikow with its ground moves, but usually latios could take care of him with surfs.

Other pokémon that could generate inconveniences did so by also altering my speed advantage, but this time because they were naturally faster (or had a Choice Scarf/Quick Claw). The most feared one was Aerodactyl as it had coverage for super effective hits against all my team members. But sometimes it had a choice band so it couldn’t change them from mon to mon, and by simply using Garchomp's sash i was out of trouble. Espeon was also sometimes faster and killed latios with shadow ball.

Then there’s other Latios and latias. Usually the male ones crumbled before my own as he is faster (but they sometimes had a Haban berry so i had to lock garchomp into outrage after latios dying) but the females proved more problematic as they were bulkier and if the Quick claw activated i was out of lack (I had one battle where it activated thrice: against Latios killing him, then brings Garchomp to the sash, survives Outrage, activates again, kills, and activates a third time to kill Raikou). No other dragon types threaten mines as we were always faster and could kill them in 1 hit.

The last big threat i can remember is Cresselia because of her bulk, recovery and lack of coverage against her. Raikou's shadow ball was the best option but by switching to him i could risk her timing the recoveries and killing me anyways. Not to mention the acces to ice beam for the dragons and well, you are forced to battle her once against palmer in the battle for the gold. Here it wasn’t the biggest problem as i usually didn’t face her as the lead (worst position to encounter her as even if i kill her she will cripple my team badly or the last member will be locked via Choice specs or Outrage).

So overall, apart from the big ice disadvange there are not a lot of big problems for the team and securing the Gold prints was very attainable with them (usually in no more than 3 attempts, 1 or 2 being the norm) but after completing that challenge and looking at my streaks, another possibility presented as i felt the 100 battles requiered for the trainer star in the tower seemed feasable. And not only that but what if, after getting that one I was to try and get at least a 100 mark on every other facility as well? That would prove that this team i barely stiched together in a whim could work better than i realized at first hand. And happily thats what i did after not so many attempts but a lot of hours overall and some crushing defeats on the higher end of the runs later, we could finally achieve it.

In the end i think this is a pretty decent team, that works pretty well for an "All Golds Run" and even for higher streaks (dont expect to go further than 100something without some luck and knowledge of the matchups though), specially if you, like me, enjoy yourself the most when using this fast paced teams that work by spamming heavy hitters.
This felt important as i was never a fan of the ultimately more consistent strategies that work slower, with bulkier pokémon and/or require some kind of long set up (although much respect to those with the knowledge to come up with them in the first place and those with the patience to execute strong streaks with them).

So this project brought me to some kind of sweetspot where, without the further need to prove myself that i can get a record worthy number I could chase other more attainable milestones while having the sense of accomplishment of seeing your ideas working out. Also in favor of that feeling was the fact that these 3 obliterated most opponents at lightning speed so most of the time they played into a kind of power fantasy, which was balanced by the fact that at any point in the streak, the wrong pokémon in the most inconvenient position, could quickly shatter said fantasy. Having gone only for Golds (as initially planned) seemed somehow easy after what i've done in the past but i found going for that 100 mark was the perfect challenge for me at this stage.

I'm ultimatrly very glad that i could walk this victory lap with the 2 legendaries that gave me the most unforgetable moments in my history with this franchise and finally give the third one of the team, Garchomp (the one that could never get to the level of what the others did), a reason to be remembered, becoming the backbone of this new powerful alliance.

So having said all of that, i will happily (although cautiously) recommend this team for anyone going for those colored prints (49 battles), and even for those who want the 100 wins in the tower or anywhere else, but keep in mind the average lenght of streak i got was around the 70 wins and after getting very familiarized with it.

If you have other options or ideas to improve the team let me know, and if you want to get what i suggested but in a faster or more consistent way look into other strategies or swap some members for better suited options like maybe scizor with the dragons, starmie as a lead and other already proven teams.

But if all you want is to try your own ideas, go ahead too as i assure you will enjoy the process and even if you fail to accomplish your initial goals, you can find enjoyment in knowing that you reached those new limits even if they pale in comparison to other old strategies. Theres no need to chase a World Record to keep you entertained (although it's still one of many ways), and as long as you keep challengin yourself while having fun, you' ve already won!


PD: Oh! and for those of you shiny hunters (as i myself used to be), i"ll throw in a couple of bonus pics.
Yeah, I was able to witness a shiny Gabite and a Sharpedo while running this team throughout all facilities. (Coincidentally both appeared on the Arcade, one in Pt while the other in HG. And i just realized they're both female sharks LoL).
I've already seen a couple in emerald, but this was the first time that this happened to me in gen 4. In the past this would have hurt a lot as i used to have the mindset of "oh no uncatchable shiny :( " but thankfully, those days are gone and now i can gleefully say "How cool is this and how lucky i am that i witnessed 2 shinies on top of the fun i was already having :)"
 

Attachments

  • 20240710_205106.jpg
    20240710_205106.jpg
    1 MB · Views: 22
  • 20240706_223118.jpg
    20240706_223118.jpg
    2.4 MB · Views: 26
  • 20240707_072702.jpg
    20240707_072702.jpg
    3.2 MB · Views: 19
  • 20240708_222552.jpg
    20240708_222552.jpg
    2.1 MB · Views: 21
  • 20240711_170741.jpg
    20240711_170741.jpg
    1.9 MB · Views: 27
  • 20240708_074922.jpg
    20240708_074922.jpg
    1.9 MB · Views: 22
  • 20240711_171304.jpg
    20240711_171304.jpg
    2.4 MB · Views: 25
Last edited:
Back
Top