OU GSC OU Viability Rankings mk. 4

Zokuru

The Stall Lord
is a Tiering Contributor
My personal take if we were to order C ranks:

[...]

The last 3 are just bad niches, Amphy being the closest to being not THAT bad. I mean, it's basically a slow Raikou, but having fire punch/dp and packing both screens is kinda good? idk.
Kazam is hard walled by too many things and encore runs out of pps too soon to be truly useful, anything MrE says about it is biased, prove me wrong.
Sandslash is............. a worse Wak that can spin?

Psychic FirePunch Toxic Recover Alakazam can force a lot of stuff into Rest and is a deadly Pokemon if you somehow manage to take Snorlax out of the picture, it's super fast and it can 3HKO and outspeed both Electrics ( Raikou is a roll after Spike tho, but any non Sleep Talk Raikou will eventually fall to Zam anyway ). Toxic + Recover makes him pretty " good " vs Tyranitar, at least it's not easy to trap. It acts a bit like Jynx, punishing hard Offensive teams, it also shares the same checks in those kind of teams , Snorlax and Vaporeon.

Machamp Cross Chop vs. Alakazam: 102-120 (32.5 - 38.3%) -- 2% chance to 3HKO after Leftovers recovery
Machamp Rock Slide vs. Alakazam: 102-121 (32.5 - 38.6%) -- 4.5% chance to 3HKO after Leftovers recovery

Even tho Alakazam bulk isn't the greatest, it's a fine check to Machamp, and it does not fear Thunder that much comparing to Jynx, even if it really doesn't want to take a para, ruining its speed.

Zapdos Thunder vs. Alakazam: 168-198 (53.6 - 63.2%) -- guaranteed 2HKO after Leftovers recovery

I think it's a super decent Pokémon that can do stuff by itself in some matchups, and prepare the field for other pokémon in others. Namely Machamp or Vaporeon.

Main Donphan niche would be the role compression of a Spinner that rekt ghost types, can check Snorlax and can Phaze. Even if Protect might be an option over Roar. I had to suffer facing it, piloted by a great player, trust be this thing has a real niche. I think C is fine for Donphan, and Alakazam should be raised in C too.
 

Oldamar999

Tien Time
I agree on the idea that ursaring should join lapras as an unranked uubl. this mon really sucks, it doesn't have any niche that isn't entirely outclassed by other normal types like Kangaskhan, Snorlax, Porygon2, and maybe even Tauros. it's higher attack stat just doesn't prevent him from being completely outclassed.
 

Royal Flush

in brazil rain
is a Past WCoP Champion
Psychic FirePunch Toxic Recover Alakazam can force a lot of stuff into Rest and is a deadly Pokemon if you somehow manage to take Snorlax out of the picture, it's super fast and it can 3HKO and outspeed both Electrics ( Raikou is a roll after Spike tho, but any non Sleep Talk Raikou will eventually fall to Zam anyway ). Toxic + Recover makes him pretty " good " vs Tyranitar, at least it's not easy to trap. It acts a bit like Jynx, punishing hard Offensive teams, it also shares the same checks in those kind of teams , Snorlax and Vaporeon.

Machamp Cross Chop vs. Alakazam: 102-120 (32.5 - 38.3%) -- 2% chance to 3HKO after Leftovers recovery
Machamp Rock Slide vs. Alakazam: 102-121 (32.5 - 38.6%) -- 4.5% chance to 3HKO after Leftovers recovery

Even tho Alakazam bulk isn't the greatest, it's a fine check to Machamp, and it does not fear Thunder that much comparing to Jynx, even if it really doesn't want to take a para, ruining its speed.

Zapdos Thunder vs. Alakazam: 168-198 (53.6 - 63.2%) -- guaranteed 2HKO after Leftovers recovery

I think it's a super decent Pokémon that can do stuff by itself in some matchups, and prepare the field for other pokémon in others. Namely Machamp or Vaporeon.

Main Donphan niche would be the role compression of a Spinner that rekt ghost types, can check Snorlax and can Phaze. Even if Protect might be an option over Roar. I had to suffer facing it, piloted by a great player, trust be this thing has a real niche. I think C is fine for Donphan, and Alakazam should be raised in C too.
Not sure about the Jynx comparison: in my book, having a sleep move and getting a clean 2hko on zapdos feels way better than... having recover. Checking Machamp is actually a good niche though, since he's something with almost no reliable checks, I can agree with you on that.

Golem is literally what you described about Donphan but better because he can boom and actually resists normal attacks. I honestly can't see taking neutral to normal as a check to Lax, especially with double-edge being the most popular pick nowadays.
 
Last edited:

Zokuru

The Stall Lord
is a Tiering Contributor
Not sure about the Jynx comparison: in my book, having a sleep move and getting a clean 2hko on zapdos feels way better than... having recover.
I only compared the way they matches-up vs common Offenses.

Golem is literally what you described about Donphan but better because he can boom and actually resists normal attacks. I honestly can't see taking neutral to normal as a check to Lax, especially with double-edge being the most popular pick nowadays.
Golem do not punish Lax clicking Curse / Drum, Donphan can check every Lax's setup set thanks to Encore. The main set would be EQ Encore Spin Roar / Protect. Where Golem just get smash'd by EQ. Donphan isn't a strudy switch to Lax as Golem can be, it's a threat forcing it to spend attacking PPs on your true walls. It prevents setup.
 

Mr.E

unban me from Discord
is a Two-Time Past SPL Champion
I have little to say, since SPL has given me absolutely no motivation for me to bother playing the game anymore.

Venusaur still deserves to chill with Meganium I think. C doesn't really need split I think though, the mons aren't really good enough to worry about the distinction but don't hate on my boy Zam. Don't underestimate the precious few hard-hitting Special attackers in this game, and Encore isn't the only support move it can carry.
 
I agree on the idea that ursaring should join lapras as an unranked uubl. this mon really sucks, it doesn't have any niche that isn't entirely outclassed by other normal types like Kangaskhan, Snorlax, Porygon2, and maybe even Tauros. it's higher attack stat just doesn't prevent him from being completely outclassed.
Speaking of Tauros, has anyone ever had any success with it in GSC? I heard Berserk Gene is hilarious on it, but i'm aware Skarmory and Cloyster give him a rough time in any case.
 

Zokuru

The Stall Lord
is a Tiering Contributor
Speaking of Tauros, has anyone ever had any success with it in GSC? I heard Berserk Gene is hilarious on it, but i'm aware Skarmory and Cloyster give him a rough time in any case.
If you play no OHKO Clause GSC you can use Double OHKO Move + Sleep Talk + Rest.

Otherwise I think it's just not good
 
Im not an expert at OU play, or pokemon in general, but i have always struggled to understand the value of Cloyster, so could someone explain to me why its rated so highly as it is?

Boom+spikes are obviously very valuable, but that aside.....
I never liked that thing in any battles i have used it, its a physical wall with ice typing, its a water type with garbage special defense, for me it sounds like the worst of both worlds.
The fact that it can die from a single non-STAB thunderbolt is just sad, even with such glaring weaknesses, its still that good?
 
Last edited:

Zokuru

The Stall Lord
is a Tiering Contributor
Im not an expert at OU play, or pokemon in general, but i have always struggled to understand the value of Cloyster, so could someone explain to me why its rated so highly as it is?

Boom+spikes are obviously very valuable, but that aside.....
I never liked that thing in any battles i have used it, its a physical wall with ice typing, its a water type with garbage special defense, for me it sounds like the worst of both worlds.
The fact that it can die from a single non-STAB thunderbolt is just sad, even with such glaring weaknesses, its still that good?
It's not supposed to wall anything, it's a pivot that boom and setup spikes.
 
With SPL and the GSC Cup coming to an end and following the successful implementation of McMeghan's approach to updating the RSE OU VR (and vapicuno's methods of analysing this) being successfully adopted for the RBY OU VR, I've conducted a survey of a subset of the GSC playing community and the results are below. Thanks to ABR, Astamatitos, BKC, Blightbringer, Century Express, Charmflash, choolio, d0nut, dice, Diophantine, Earthworm, Excal, Fear, Flares., FriendOfMrGolem120, HSA, Hyogafodex, Jimmy Turtwig, Jorgen, k3nan, Lavos, Lojh, M Dragon, McMeghan, Mr.E, Sceptross, sulcata, and Zokuru for their contributions.

Before I get to the actual VR, some details on the process I used to come to the below rankings:
Collection of data
- I set up a Google form with each Pokemon from the old VR having their own question asking the user to enter the ranking number for that Pokemon.
- Only the first 35 questions were mandatory (up to the end of the old B rank + Alakazam).
- I added questions for Jumpluff and Lanturn to the end.
- There was also a free text answer box at the end where I asked if there were any Pokemon that should appear on the VR that don't already.
- Where someone accidentally put in the same numbers twice, I contacted them and asked them what they intended, then corrected this in the data.
- I provided the link to the form to GSC players on Discord and also some on the Smogon forums if I didn't have their Discord tag.
- I allowed anyone who had submitted their own rankings to view others' rankings.
- I opened a discussion channel to anyone who had submitted rankings to discuss their rankings.
- I allowed players to submit edit requests to their rankings and then made these changes in the data.
- I left the form open for about two weeks.

Formation of rankings
- Following the method McMeghan used for his RSE OU rankings, I removed the outliers from the rankings before forming the rankings. This did not affect the rankings, aside from breaking ties.
- I then used the average ranking as the method to determine which Pokemon was higher rank.
- There were two ties after this. They were resolved by the using the average of the rankings with outliers included.
- I did not allow any Pokemon to be ranked higher than any Pokemon with a higher number of rankings than itself. e.g. Since Blastoise was only ranked by two people, it could only be ranked higher than other Pokemon that were ranked only twice.
- I did not remove outliers for any Pokemon that was ranked fewer than six times.
- I required a Pokemon to have been ranked at least twice to be included on the list at all.
- To split the Pokemon into tiers and sub-tiers, I used the method vapicuno created where I looked at a graph of the average of the rankings plotted on a graph with their standard deviations, then eyeballed it with some input from vapicuno and some of my own discretion (on where to put Rhydon, Tentacruel, Donphan, etc).
- It should be noted that once we get to C ranks and lower, it becomes more difficult to distinguish sub-tiers due to the high standard deviations (which represent wildly different rankings of each Pokemon).

Flaws
- The method used to collect the rankings was quite awkward (the google form). It almost certainly caused a degree of bias due to people's desire to finish quickly and having to recall the numbers they had already entered. I'll look into alternatives if I ever do this again.
- There were some cases where people ranked things the same rank accidentally. I contacted most of them and asked for their intended ranks, but there was at least one I didn't catch (k3nan's ranking of Mr. Mime, which was "somewhere between 35 and 40". I treated this as 39.9 in the image rankings.
- The ability to meaningfully define a border for ranks below B based on averages + standard deviations is very limited. The divisions are therefore quite arbitrary.
- I would consider introducing the ability to vote to exclude certain Pokemon from the rankings if the player doesn't consider them viable in future iterations.

Without further ado...

Updated VR

S+ Rank
01
Snorlax
S Rank
02
Zapdos
S- Rank
03
Raikou
04
Cloyster

A+ Rank
05
Exeggutor
06
Gengar
07
Skarmory
08
Steelix
09
Tyranitar
A Rank
10
Machamp
11
Vaporeon
12
Nidoking
A- Rank
13
Forretress
14
Suicune
15
Marowak

B+ Rank
16
Starmie
17
Jynx
18
Misdreavus
19
Umbreon
B Rank
20
Rhydon
21
Heracross
22
Golem
23
Miltank
Tentacruel Rank
Note: Tentacruel was decidedly below the above rank but also decidedly above the below ranks in a large number of rankings, even having an unusually low standard deviation for the B ranks, so it has been given its own rank between B and B-.
24
Tentacruel
B- Rank
25
Blissey
26
Charizard
27
Espeon
28
Dragonite

C+ Rank
29
Porygon2
30
Jolteon
31
Smeargle
32
Alakazam
33
Houndoom
34
Muk
C Rank
35
Clefable
36
Moltres
37
Quagsire
38
Meganium
39
Piloswine
40
Kangaskhan
41
Donphan
C- Rank
42
Ampharos
43
Sandslash
44
Scizor
45
Entei
46
Kingdra
47
Shuckle
48
Aerodactyl
49
Gligar
50
Articuno

D Rank
51
Nidoqueen
52
Omastar
53
Qwilfish
54
Jumpluff
55
Pikachu
56
Venusaur
57
Slowbro
58
Poliwrath
59
Typhlosion
60
Tauros
61
Ursaring
62
Raichu
63
Electabuzz
64
Lanturn
65
Crobat
66
Mr. Mime
67
Blastoise
68
Hitmonlee
69
Slowking


Visual representations of results

Here is a visual representation of everyone's ranks, as well as the old rankings and the new rankings. In the New VR column, I have formatted it so that green indicates a rise and red indicates a fall when compared to the old VR. Blue indicates no movement. The pale green and red in everyone's rankings indicate what they ranked higher (green) and lower (red) than the new VR.


Here is the graph with standard deviations plotted on the averages (note- the ranks appear to the right of the actual point on the plot). This was used to decide the sub-tiers above. Thanks vapicuno for assisting with this:


Here is a graph of all the averages only, with the Pokemon names listed against the points on the plot:


How the tiers are formed from the plot of average ranks and standard deviations


Basically, all Pokemon within a tier should be roughly a standard deviation at most from each other. Tiers 1 and 2 (S+ and S) are very well defined with little to no deviation, i.e. almost everyone thought that Snorlax is #1 and Zapdos is #2. Several people thought that Cloyster should be ranked above Raikou, which led to their standard deviations overlapping on the plot. Therefore, they are tiered together in S- in this new VR.



Ranks 5 through 9 are also all clearly within a standard deviation of each other, and relatively few people ranked anything outside of the 5-9 group between ranks 5 and 9 (only the very occasional Machamp or Nidoking). It was therefore a no-brainer to make them the A+ tier.



The next clear group is group A, the Machamp Vaporeon Nidoking group. Machamp and Vaporeon both had relatively low standard deviation, while Nidoking's ranking fluctuated considerably more, however they all averaged out in similar fashion and this group can therefore be fairly objectively split from the following groups.



Next was the Forretress Suicune Marowak group, which was again quite clearly distinct from the groups above and below it (all within 1 point in terms of average rank). This group therefore became A-.



B+ contains 16 through 19. All of these Pokemon are within a standard deviation of Starmie's average, although Umbreon is a decent gap away from Misdreavus. This group is fairly well defined, although Umbreon is a bit of an outsider, being exactly one point away from Misdreavus in terms of average rank.



B rank is where groups start to encompass larger gaps and lose their well-definedness. Standard deviations also begin to rise above 2, with only Rhydon being under 2. Tentacruel was more than 2 points below Miltank but exactly 2 points above Blissey -- not fitting in either group very well at all. It therefore has its own special rank under B.



B- has the most controversial Pokemon yet in Espeon, the first Pokemon to have a standard deviation greater than 3. Charizard also had a remarkably low standard deviation for a Pokemon in the low B ranks. Dragonite closes out the B- rank, since there is a fairly large gap between itself and Porygon2, which is well outside of a standard deviation from Blissey.



C+ continues the trend of increased variance in standard deviations and contains the most Pokemon yet (6). A full standard deviation from Porygon2 gave us Jolteon, Smeargle, Alakazam, Houndoom, and Muk in this tier.



Clefable is the first in C rank, a tier that continues to list viable but only situationally effective Pokemon. It was becoming harder to define ends of the tiers based purely on standard deviation and average rank, so I had to make a judgement call on where the end of the tier would be and I chose to include all Pokemon that had an average rank of below 40, which just barely included Donphan at rank 41.




C- contains some potentially underrated Pokemon with a lot of unique qualities and typings that happened to be all grouped together in the average ranks. I again had to make a fairly arbitrary judgement call and decided to include everything between average 40 and average 50 in this tier. D tier, then, contains the rest -- mostly Pokemon you wouldn't consider using in OU unless your handle is Conflict or Lavos.


Observations
Changes when compared to the old VR that I found particularly notable include:
-
Cloyster (-) closing in on Raikou as a contender for #3. The tier is continuing to grow more and more Spikes-centric, and Ground-types are also continuing to rise in popularity across the board, with the exception of Marowak. Several notable players such as BKC, FriendOfMrGolem120, and ABR, as well as veteran Jorgen rated Cloyster as their #3. Lojh is crazy though.
-
Gengar (+2) shot up the ranks due primarily to its role as the tier's most flexible spinblocker. With the increase in Starmie and Golem recently, having a backup plan for when Cloyster has been unexpectedly murdered after laying Spikes gives some teams a way to keep up pressure and allows them to keep more defensive teams on their toes. Gengar was incredibly close to taking the #5 position from Exeggutor this time around and I would not be surprised if it manages it next time.
-
Machamp (+3) has jumped up above Vaporeon, Marowak, and Suicune to round out the top ten. I suspect this is partly due to previous bias against its reliance on 80 acc Cross Chop. Exeggutor has also been seeing less play lately for reasons I'll get into shortly, which has resulted in some of Machamp's coverage woes being alleviated somewhat.
-
Nidoking (+2) has seen massive increases in usage lately, especially Thief variants. This has corresponded with a jump of two positions in the VR update. Its role of stallbreaker is highly valuable, and it is one of the hardest Pokemon to truly wall in the tier. Its knack for duelling with Zapdos also does it huge favours. Some players even ranked Nidoking in the top nine, namely M Dragon, Zokuru, and k3nan.
-
Marowak (-4) was one of the hardest hit Pokemon in these rankings, perhaps due to its poor synergy with the currently trending Pokemon. While Rhydon's increase in popularity was a good thing for Marowak due to Marowak outspeeding it, it speed ties with Golem, which can also explode on it.
-
Jynx (+3) continues to maintain its popularity gained through its dominating matchup against offense mainstays such as Cloyster, Zapdos, and Exeggutor, preying upon the limited counters most teams have against it. Jynx is almost certainly responsible for Exeggutor almost relinquishing its #5 spot to Gengar thanks to its absolutely crushing matchup against Eggy. Jynx saw BKC rate it as high as 11th, with others such as Diophantine and Excal also rating her highly. Time will tell if Jynx continues to rise beyond its current rank of 17.
-
Golem (+2) also continues to rise, albeit by less than some may have expected. FriendOfMrGolem120 was not even the highest to rank Golem at his ranking of 15 -- ABR ranked Golem as high as 13. Golem's niche as a Rapid Spinner, Normal + Fire resist, phazer, and exploder all in one compact shell is what has driven its unprecedented rise in popularity in such an old gen. I suspect it would have risen even higher were it not for the format of the form somewhat encouraging people to rank Pokemon in the Bs in a similar order to the original VR.
-
Miltank (-5) has dropped a remarkable distance in this update, having seen very low usage for a long period. It seems to be having a resurgence thanks to McMeghan lately though. It has dropped in favour of Rhydon, Heracross, and Golem. It actually tied Golem exactly in terms of average rank -- Golem only took the higher rank thanks to its higher ranking with outliers included.

I look forward to hearing what others have to say about the changes.

Ranking Correlation
Here is a ranking correlation graph (S and A ranks). This is another method of showing how similarly different people ranked Pokemon in A rank. Lighter is more similar and darker is less similar. For more info check vapicuno's post here. Thanks to vapicuno for making this and the other correlation graphs.

- Some of the most similar rankings were those between Century Express + M Dragon and McMeghan + HSA.


- Meanwhile, Mr.E seems to disagree with just about everyone. Which is odd, considering he was the only one to get the first 7 Pokemon in the same order as the final outcome.

Here is a correlation graph for the B ranks.

- Here, it seems like BKC and choolio have many disagreements, as do Mr.E and Zokuru.
- The two most similar B ranks were Hyogafodex and Charmflash:


Lastly, a graph for S to B-, from ranks 1 to 28.

- Again, Mr.E disagrees with almost everyone.
- The most similar rankings look to be those of dice and d0nut.


I hope everyone enjoyed looking at the results and I'm sorry about the wait. The data is available in spreadsheet form here.
 
Last edited:

Jorgen

World's Strongest Fairy
is a Forum Moderator Alumnusis a Community Contributor Alumnusis a Contributor Alumnusis a Past SPL Champion
Great work collecting these data. I might play with them a bit myself. In particular, I'm interested to see how finely or coarsely grained one can cluster the range of opinions about GSC. Are there two "camps", for instance?

EDIT:
I ran hierarchical clustering to sort the correlation heatmaps and found evidence for different "camps" (shown as different "squares"). These correlations are assessed for all 35 rankings of all the mandatory Pokemon:
184841

The Fear-Dio-BKC-ABR group seems to comprise the extrema of one group, whereas the Flares-choolio-d0nut (and to a lesser extent, dice-Charmaflash) group seems to act as the cohesive standard-bearer for the other. These "camps" seem to overlap: most notably, the Lavos-FOMG-Excal-hsa-Earthworm nexus seems to straddle both "squares". There's ever finer structure that you could squint & possibly find in this heatmap, but I focused on the big thing that popped out to me immediately.

Despite using hierarchical clustering to find this structure, this is the point where I stop leaning on it to group players together. We clearly have groups that overlap substantially, yet if I were to move forward and build a dendrogram, for instance, it would force the overlapping bloc to side with one cluster or another. I'm going to trust my eyeballs on this one and try to find which Pokemon these two "camps" disagree most on.

For starters, I'm going to take a hierarchical clustering approach to tiering as a whole. First, I need a dissimilarity matrix to define distances between Pokemon, then I need to run the clustering algorithm to sort the Pokemon. The dissimilarity matrix is shown below, sorted by their ranking obtained from averaging the ranks (i.e., prior to any clustering algorithm). For those who would have me show my work: the dissimilarity rating is obtained by computing, for each pair, the rate at which Pokemon X is ranked better than Pokemon Y, then applying the logit transform and taking the absolute value of that.
184855

You may notice "squares" here, as well. Whereas for players we may call them "camps", here we might call them "tiers" because we can infer an ordinal relationship among the clusters. If I were to group the VRs myself:

We see that there's a sharp delineation between Snorlax (what I'll call S) & Zapdos/Raikou/Cloyster (A+) tiers
then another marking the Egg/Gengar/Skarm/Steelix/Ttar (A) tier
then another marking Champ/Vap/Nido (A-)
then another still marking Forretress/Suicune/Marowak (B+)
From Jynx onward, things get hairier, as there's more overlap between tiers:
I might group all of Starmie/Jynx/Misdreavus//Umbreon/Rhydon together (B)
then Heracross/Golem/Miltank (B-), and perhaps Tentacruel also goes in B- to avoid silly things like "Tentacruel tier", even though it only partially overlaps with this tier.
From here, Blissey/Charizard/Espeon/Dragonite belong together (C+), although Espy & Dnite seem to overlap a decent bit with the tier below, and Blissey with the one above.
then Pgon2/Jolteon/Smeargle/Alakazam/Clefable/Muk (C), although a case could be made for breaking Pgon2 & Jolteon away from the rest of this group.
Kangaskhan signals the beginning of the even-lower tiers.

As a caveat, I only ran this on the 35 mandatory Pokemon again, but it could totally be done for all Pokemon. I just lost steam & wasn't inspired to whip up the code needed to figure out where Houndoom & Qwilfish belonged in the grand scheme of things.

Of course, there's using my eyeballs, and there's using the heatmap & dendrogram that agglomerative clustering gives us & picking a threshold that gives reasonable tiers:
184868

184871

Here, all colors but blue mark groupings into tiers. You may also notice that the order got shuffled a bit; hierarchical clustering (at least in its base form) isn't so good at maintaining ordinal relationships. The Y axis marks distance, with the horizontal lines merging leaves together indicating how far apart two leaves are. Coloring was determined by setting an arbitrary (but "reasonable-looking") threshold, where any leaves that are less than 1.25 apart are joined & colored. We see a broadly consistent tiering pattern, agreeing with the rank-averaging decision to tier Zapdos separately from Cloyster/Raikou and the idea of "Tentacruel tier". It's not a huge fan of too many distinctions among lower tiers though; it really wants to lump Zard/Bliss/Pgon2/Jolt/Dnite/Espeon into one big pot. Left up to agglomerative clustering, then, the VRs might look like:
S+: Snorlax
S: Zapdos
S-: Raikou/Cloyster
A+: Egg/Gengar/Skarm/Steelix/Ttar
A: Champ/Vap/Nido
A-: Forr/Suicune/Marowak
B+: Starmie/Jynx/Missy/Umbreon
B: Rhydon/Heracross/Golem/Miltank
Tentacruel: Tentacruel
B-: Blissey/Charizard/Espeon/Dragonite/Pgon2/Jolteon
C: Smeargle/Alakazam/Clefable/Muk/Kanga

All in all, multiple methods give us VRs similar to what averaging ranks does. Which is good and points to how "objective" the VR really are (at least up until roughly Tentacruel it seems, at which point putative "tiers" begin to overlap too much to draw meaningful distinctions).

So, okay. Now I've run my own personal VR calculations. BUT I still haven't addressed the "camps" from earlier. Well, my plan all along was to introduce my VR method (at least the first bit, where I compute head-to-head "better-than" rates), then compute them separately for the two different "camps" to see what differences emerge between the two. In particular, I'm looking for differences between the "extrema" of the first bloc (ABR/BKC/Dio/Fear) and the "standard bearers" of the second bloc (Charmaflash/dice/d0nut/choolio/Flares). Although Lavos/FOMG/Excal/hsa/Earthworm form a compact group, I'm excluding them because they "straddle" both clusters.

Here's the heatmap showing the difference between the two. To help orient you: red means that bloc 1 prefers pokemon X. Blue means that bloc 2 prefers pokemon X.
184880

We see a good deal of disagreement is about the mid-lower ranks: standout rows/columns include Jynx, Golem, and Starmie. To attach a single number to each Pokemon, I simply compute the mean difference across each row. I make the choice to first remove all zeros before computing the average; otherwise, the fact that everyone agrees Snorlax is better than Starmie will muddle our effects.
184881

Here, +ve means bloc 1 likes the Pokemon (ABR/BKC/Dio/Fear), and -ve means bloc 2 (Charmaflash/dice/d0nut/choolio/Flares) likes it. We see that the Pokemon with +ve ratings beyond 1 standard deviation of 0 are generally offensive in nature, or are very good against offensive teams: Jynx, Alakazam, Gengar, Cloyster, Porygon2, and Zapdos. Meanwhile, Pokemon with -ve ratings beyond 1 s.d. of 0 are generally stally in nature, or very good against traditional stally teams: Muk, Suicune, Miltank, Marowak, Raikou, Skarmory, and Exeggutor. It seems, then, that we do have evidence for a stall-offense spectrum in GSC opinion. Or, perhaps another way to think of it is "old-school" vs. "new-school", as the "cutting-edge" seems to have acquired a taste for Jynx and Cloyster and a distaste for Miltank, Marowak, and Skarmory.

One caveat here: Note in the first heatmap how there's loads of players that don't really fit into any "cluster". The formation of stall and offense "camps" is ultimately a second-order organizing principle to "consensus vs. weirdo". When I sort the heatmap not by agglomerative clustering, but simply by computing the mean pairwise correlation of each player with all other players, I get something that looks like this:
184883

Here, we see a trend that captures all players, and not just a subset that are pre-disposed to agree with one another. Namely: we have a spectrum of players that are simply more or less inclined to agree with some "consensus". Simply plotting that mean "agreeability" and subtracting out the grand mean correlation, we find the following:
184884

Most people seem to vary smoothly in terms of how well they agree or disagree with other folks, with two big exceptions: Mr.E and, to a lesser extent, myself. We both disagree with the consensus beyond 2 standard deviations of the mean, making us massive weirdos. For what it's worth, I think Mr.E's ridiculous Ttar ranking did him in. As for myself; I suppose my appetite for Vaporeon and relative distaste for Gengar & Exeggutor did it?

Despite everyone else falling within 1 s.d. of the grand mean, "agreeability" still gives us a way to describe players like Lojh and Sulcata (not to mention McMeghan) who aren't massive weirdos (or even weirdos at all) but nonetheless didn't fit neatly into either of the clusters from earlier. As such, I'm inclined to think of the less-satisfying "outsider-insider" spectrum as the main one explaining GSC opinions, with "offense-defense" being a secondary spectrum.
 
Last edited:

Mr.E

unban me from Discord
is a Two-Time Past SPL Champion
dang I really am the best, I got the most Top X same over anyone else (Funny enough, I'm the only one with literally zero matches below that.)

I know it's too late to fix it now, but it'd be nice if you had used a darker shade of color for bigger outliers, so people could easily find them for their own amusement.

  • shoutouts to my man sulcata for being the only other bro to rank blastoise
  • Jorgen didn't even rank ttar the highest wtf noob. I also respect the #2 Cloyster pick from Lojh as well.
  • I don't respect Zokuru and Flares. for putting Raikou #2. :pikuh: Although the former is the only other person besides myself who ranked Ttar out of even the Top 10.
  • dis nigga choolio literally ranked 72 pokemon
  • meganium is 38 but venusaur is 56 and they're like the same pokemon except fighting (machamp) vs ground (marowak) resist and Venusaur gets Sleep Powder :pikuh:
  • golem is a shit spinner stop fucking using it as one jesus christ you can't be a good spinner when the spiker can literally switch in and force you out guys stop it :blobstop:
  • ... Honestly why do you people have such a boner for Charizard anyway
Most people seem to vary smoothly in terms of how well they agree or disagree with other folks, with two big exceptions: Mr.E and, to a lesser extent, myself. We both disagree with the consensus beyond 2 standard deviations of the mean, making us massive weirdos. For what it's worth, I think Mr.E's ridiculous Ttar ranking did him in. As for myself; I suppose my appetite for Vaporeon and relative distaste for Gengar & Exeggutor did it?
How bad does it look for me if you move Ttar up, say, 10 spots and move everything else down 1 to compensate? I still have a lot more relative outliers: +7 Marowak / Blissey, +5 Rhydon / Heracross, -8 Golem / Tentacruel if they're important enough to matter. (Well, they do matter more than Blissey at least.) If a whole +/- 4 on Vap and Eggy place you in the hallowed "disagreeable" camp, I imagine I'm still quite disagreeable even with an alleged reasonable Ttar ranking.

While I'm here, allow me to make the case against Ttar, in brief: It can't switch into fucking anything except it's not as dangerous offensively as Machamp who suffers similarly, or Marowak who lacks Leftovers but can eat Thunders. (Ttar, incidentally, is like the one thing Machamp switches into for free.) Its only utility defensively is vs Normal+Fire Lax and even then you still need a defensive variant with Rest or it gets destroyed by status and Spikes/chip, at which point it's not an offensive threat anymore. I don't even like Rhydon and I still think it compares favorably vs Ttar if you specifically want a Rock-type for Firelax resist. Pursuit is cool and all but STAB on Earthquake and Electric immunity are much bigger and broadly applicable traits.
 
Last edited:
[*]dis nigga choolio literally ranked 72 pokemon
Dude's missing like 10 UU mons. He really ought to up his game.

While I'm here, allow me to make the case against Ttar, in brief: It can't switch into fucking anything except it's not as dangerous offensively as Machamp who suffers similarly, or Marowak who lacks Leftovers but can eat Thunders. (Ttar, incidentally, is like the one thing Machamp switches into for free.) Its only utility defensively is vs Normal+Fire Lax and even then you still need a defensive variant with Rest or it gets destroyed by status and Spikes/chip, at which point it's not an offensive threat anymore. I don't even like Rhydon and I still think it compares favorably vs Ttar if you specifically want a Rock-type for Firelax resist. Pursuit is cool and all but STAB on Earthquake and Electric immunity are much bigger and broadly applicable traits.
Tyranitar is like the Kobe Bryant of GSC. You never know what you're going to get.
 

Mr.E

unban me from Discord
is a Two-Time Past SPL Champion
I can't say I looked through the entire list and checked to see if anyone was missing anything notable in the UU space. The only mons actually tagged as UU in the site Dex that unequivocably deserves to be ranked is Quagsire and maybe Ampharos, plus Chansey/Nidoqueen if you're talking things that might see play if they weren't 100% redundant to existing OU options.

Also I actually like Golem, I just think it's bad as a Spinner. Explosion is much better than it was in RBY, so it's an appealing enough general attacker. I should have ranked it higher but I semi-intentionally lowballed it because I knew people would be rating it higher than usual thanks to the recent explosion :mehowth: of its popularity as a Spinner.
 
u use golem in conjunction with cloyster to tox/boom in the ditto to open for the spin later in the game. far less hard to set up than u make it seem--cloy isn't exactly the most shielded pkmn. and for a firelax phazer it's good to have. combination of a lot of defensive utility that you're underselling imo.
 

Mr.E

unban me from Discord
is a Two-Time Past SPL Champion
All that says to me is that people are too laissez faire with their Cloysters and that the metagame should adjust in due time if Golem remains so popular as a Spinner. Or maybe we'll at least see more stupid shit to take advantage of it, like more Surf Snorlax nonsense kinda like Thunder becoming increasingly popular on Lax just to dick with people gut switching Cloyster in to get Spikes up ASAP. (DE + Spikes + Surf KOs Golem.) Honestly might be a better third attack for DE+Thunder Lax than Earthquake anyway... Also Golem still loses to meme Giga Drain Forry, which does tend to stick around if only because Forretress is a dumb pokemon with virtually no other utility so it doesn't have much other reason to switch in. Don't tell me anyone is a bad enough dude to run Fire Blast over Explosion (or Roar).

Compared to Starmie which you can just hard into vs Cloyster and there's little the Cloyster player can do about it if they don't have Thunder Misdreavus backing it up, having to finesse an opportunity for Golem to Spin actually is a real downside though. Golem has to find an opening later, after who knows how much Spikes damage already accumulates, while Starmie and to a lesser extent more dedicated Cloyster/Forry sets never allow Spikes to stay up at all.

The escalating Spikes war in GSC is kinda silly anyway. It's one fucking layer, dudes, on top of the much lower amount of other passive damage available in the game than succeeding generations. There are more games to be won blowing Cloyster up on Starmie and running a train on the remaining five with Machamp than fucking around with Toxic trying to get Spikes to stick after 100 turns.
 
I've now implemented the rankings in the OP of the thread with the permission of Lavos. I also decided to put Tentacruel in the nearest group rather than having it in its own rank after receiving some feedback.

Also, great stuff Jorgen (if anyone didn't notice he edited a crapton of cool stats into his post).

Outliers
By request, here is some data which highlights the outliers (S+ to B-) by taking the absolute value of the difference between the given and average ranks. All of this is based on the data with outliers included.


Same data but showing where Pokemon were ranked higher than average and lower than average rather than just focusing on the biggest difference:



It's no surprise given some of the ranking correlation graphs that Mr.E is responsible for some of the biggest outliers in the data, namely his ranking of Tyranitar at #21 (average rank was 8.5) and Charizard at #45 (average rank 27.29). His rankings of Dragonite (#39), Blissey (#18), and Tentacruel (#32) were also quite far from the average. Players such as Lojh() and k3nan thought Blissey was a full 16 ranks worse than Mr.E's #18, giving it a ranking of #34.


Aside from Mr.E, Zokuru had the biggest outlier with Espeon at #42 (average rank was 28.607). I find this a little surprising because there have been instances recently where Espeon has had good showings, such as in the GSC Cup. The rise of Nidoking and Gengar should also help Espeon succeed in theory (at least if it can win speed ties and avoid getting paralysed by Zapdos). Another player who ranked Espeon surprisingly low was Diophantine at #36.


The next biggest outlier was ABR's Golem ranking, which was a 8.85 ranks higher than average at 13. Golem is a relatively new feature in the metagame and I expect it will continue to climb the ranks as more and more people realise its potential. ABR is potentially ahead of the curve here, but it remains to be seen if he has gone too far. I suppose that could change if we see more SurfTar and SurfLax though. FriendOfMrGolem120 had a similarly high Golem ranking at 15. The lowest ranking was choolio's at 27.


Aside from Mr.E's Tyranitar, some of the biggest outliers within the A rankings were Astamatitos's Suicune and Marowak rankings and BKC's Marowak ranking. Asta really doesn't think Suicune is good, rating it at #21 and Marowak at #20, when their average ranks were just over 14. BKC had the lowest Marowak ranking of anyone at 21 -- he must value his Leftovers. The highest ranking for Marowak was Mr.E at #8, which makes sense considering his refusal to believe that one needs to place heavy importance on the Spikes game to succeed. dice rated it the next highest at #11. Flares. and sulcata both ranked Suicune as their #10, which is a very high rating for a Pokemon that offers a relatively low cost/risk switch-in opportunity to all of the top four Pokemon in the metagame in my opinion.


Looking even higher, there were some fairly sizeable outliers even among the A+ ranks. Jorgen, Blightbringer, and Charmflash all ranked Steelix far lower than most at #12 when it averaged 7.714 and even saw McMeghan, sulcata, and d0nut rank it as their #5. I would be very interested to hear the justification for the low rankings considering it had the best win rate in the top 10 most used Pokemon in both SPL 9 and SPL 10. We are even beginning to see successful deviation from the norm with occasional Rest Steelix as well as DBreath / BSlam Steelix.


Finally, Fear was the only player to rank Skarmory outside the top 10, ranking it #11 when its average rank was just under 7. Of note is that the lowest rank anyone else gave it was #9, and even then, only Astamatitos and Diophantine ranked it that low. Fear certainly trends towards offense more than he does stall, and likely hates the passivity, but even others who have complained about Skarmory in the past have warmed up to it and acknowledged its strengths to some extent in this ranking exercise.

If anyone cares about the outliers below B-, they're here:
 
I've now implemented the rankings in the OP of the thread with the permission of Lavos. I also decided to put Tentacruel in the nearest group rather than having it in its own rank after receiving some feedback.

Also, great stuff Jorgen (if anyone didn't notice he edited a crapton of cool stats into his post).

Outliers
By request, here is some data which highlights the outliers (S+ to B-) by taking the absolute value of the difference between the given and average ranks. All of this is based on the data with outliers included.


Same data but showing where Pokemon were ranked higher than average and lower than average rather than just focusing on the biggest difference:



It's no surprise given some of the ranking correlation graphs that Mr.E is responsible for some of the biggest outliers in the data, namely his ranking of Tyranitar at #21 (average rank was 8.5) and Charizard at #45 (average rank 27.29). His rankings of Dragonite (#39), Blissey (#18), and Tentacruel (#32) were also quite far from the average. Players such as Lojh() and k3nan thought Blissey was a full 16 ranks worse than Mr.E's #18, giving it a ranking of #34.


Aside from Mr.E, Zokuru had the biggest outlier with Espeon at #42 (average rank was 28.607). I find this a little surprising because there have been instances recently where Espeon has had good showings, such as in the GSC Cup. The rise of Nidoking and Gengar should also help Espeon succeed in theory (at least if it can win speed ties and avoid getting paralysed by Zapdos). Another player who ranked Espeon surprisingly low was Diophantine at #36.


The next biggest outlier was ABR's Golem ranking, which was a 8.85 ranks higher than average at 13. Golem is a relatively new feature in the metagame and I expect it will continue to climb the ranks as more and more people realise its potential. ABR is potentially ahead of the curve here, but it remains to be seen if he has gone too far. I suppose that could change if we see more SurfTar and SurfLax though. FriendOfMrGolem120 had a similarly high Golem ranking at 15. The lowest ranking was choolio's at 27.


Aside from Mr.E's Tyranitar, some of the biggest outliers within the A rankings were Astamatitos's Suicune and Marowak rankings and BKC's Marowak ranking. Asta really doesn't think Suicune is good, rating it at #21 and Marowak at #20, when their average ranks were just over 14. BKC had the lowest Marowak ranking of anyone at 21 -- he must value his Leftovers. The highest ranking for Marowak was Mr.E at #8, which makes sense considering his refusal to believe that one needs to place heavy importance on the Spikes game to succeed. dice rated it the next highest at #11. Flares. and sulcata both ranked Suicune as their #10, which is a very high rating for a Pokemon that offers a relatively low cost/risk switch-in opportunity to all of the top four Pokemon in the metagame in my opinion.


Looking even higher, there were some fairly sizeable outliers even among the A+ ranks. Jorgen, Blightbringer, and Charmflash all ranked Steelix far lower than most at #12 when it averaged 7.714 and even saw McMeghan, sulcata, and d0nut rank it as their #5. I would be very interested to hear the justification for the low rankings considering it had the best win rate in the top 10 most used Pokemon in both SPL 9 and SPL 10. We are even beginning to see successful deviation from the norm with occasional Rest Steelix as well as DBreath / BSlam Steelix.


Finally, Fear was the only player to rank Skarmory outside the top 10, ranking it #11 when its average rank was just under 7. Of note is that the lowest rank anyone else gave it was #9, and even then, only Astamatitos and Diophantine ranked it that low. Fear certainly trends towards offense more than he does stall, and likely hates the passivity, but even others who have complained about Skarmory in the past have warmed up to it and acknowledged its strengths to some extent in this ranking exercise.

If anyone cares about the outliers below B-, they're here:
Another great post that costed me 30 minutes of my life, good work EW. I can't speak for everyone but my reasoning for placing Steelix so low is its unreliability. It is a mon that can do everything in theory, it is supposed to switch into Electrics, check Lax and break via Roar into eventually Boom but it can fail at every single role. It is the only Lax check weak to both Fire- and Ground-coverage (even weak to the occasional Surf) while also hating Lovely Kiss. Admittedly, it does not take a whole lot from EQ, however, with a lack of recovery, its inability to even outspeed Snorlax and Spikes being up constantly against most of its common team structures, it's not going to be around for long versus that set either. Against Electrics, you're always in danger of running into Hidden Power Water, which can 2hko you on the spot if Spikes are up. Offensively you're usually relying on a boom trade and Roar RNG to get anything done, and both can go incredibly wrong, especially vs. Water-types (who I rate quite highly, which is probably also why Nidoking is lower than average for me) and Rapid Spin users.

To summarize
- Unreliable vs. multiple Lax sets
- Unreliable vs. Electrics
- Unreliable offensively due to its over-reliance on boom
- Unreliable due to Roar RNG in the midgame
- Invites Spikes
- Team structures oftentimes lack Rapid Spin
- Needs Spikes to function but loses to every Rapid Spin user to some extent
- I rate all the Waters quite highly

Despite all this, I still think that Steelix is a good Pokemon in the GSC OU metagame, otherwise I would not have ranked it in the top 12. I just listed the negative aspects of Steelix, the reasons why I ranked it lower than other people.
 

Jorgen

World's Strongest Fairy
is a Forum Moderator Alumnusis a Community Contributor Alumnusis a Contributor Alumnusis a Past SPL Champion
Admittedly I've been out of the tourney scene for a while, so I didn't know about Steelix's recent winrate surge or why that may have been the case (although fwiw I always found those winrate stats to be awfully noisy).

Regardless, my personal distaste for Steelix arises from how toothless it is. Unless you blow it up, it doesn't really contribute offensively, and teams frequently look to the Ground-type slot to generate offense. Defensively, it can vacillate between unbreakable lynchpin and easily worn-down stopgap. For something as offensively lackluster as Steelix, that sort of variance in defensive performance is quite unappealing. Plus there's a number of rising trends in the current metagame (heavier Jynx usage, Thief Misdreavus & Gengar, stronger emphasis on Spikes and on Cloyster in particular) that are rather unfriendly to Steelix. Finally, my personal tastes admittedly color my judgement: I find it often competes with Tyranitar for a teamslot (mostly because it's very easy to have 4+ fighting-weak Pokemon if you run both on the same team), and I generally prefer Tyranitar's versatility and ability to support the team with Pursuit over Steelix's bland defensive stylings.

That said, that diatribe is mostly justifying my relative ranking of Steelix. It's still a very good Pokemon in an absolute sense: It checks loads of boxes on defense, has access to Explosion which is always a game-changer, and can be essentially invincible in certain endgames. It's definitively A-rank, and I can even sympathize with those who put it at 7 or 8. Those who ranked it as high as #5 seem a little off their rockers in my eyes, though (any of you Steelix apologists gonna take that lying down?!?).

Also, for those of you using Dragonbreath Steelix: If you aren't already, I suggest using minimum SpA. That way, you do so little damage to Zapdos that its Leftovers heal it back up to full, meaning it can't immediately use Rest to erase paralysis.

EDIT: for the last point, I did indeed overlook the Special DV link. However, I believe PS! does offer you the ability to adjust stat experience (implemented as EVs) to lower your SpA stat enough without having to touch your DVs. If it doesn't, then fully disregard everything about Steelix and SpA.
 
Last edited:
At full and mid health Steelix is a more useful normal resist than Tyranitar for the types of offensive teams that tend to run them.
It's just got great role compression in that it can do a better job in checking both non fire lax and electrics at full health, and then at middle health the mind games of explosion are always there.
Compared to Skarmory it actually has offensive presence, and can pressure with curse, explosion, and the ability to spread para. Skarm is great as a pure wall, but that's where it's job ends so that's definitely not going to suit more offensive teams.
It's hard to compare it to the other 2 mons in its rank of Exeggutor and Gengar, especially when they both are often on the same sorts of teams. What I can say is that on explosion offense Steelix is almost invariably there, whereas you often pick and choose between the other two. Steelix's role is that essential and irreplaceable.
 
Also, for those of you using Dragonbreath Steelix: If you aren't already, I suggest using minimum SpA. That way, you do so little damage to Zapdos that its Leftovers heal it back up to full, meaning it can't immediately use Rest to erase paralysis.
Jumping in just to mention for the people who are not aware of every mechanic yet the generation has to offer: SpA and SpDef share the same DV and EV Values, thus decreasing SpA decreases SpDef so you would also make your steelix take more from moves from the special side. So it would be wise to not set your SpA to minimum because you would set your SpDef to minimum as well.
 

Mr.E

unban me from Discord
is a Two-Time Past SPL Champion
My love for Marowak isn't so much about Spikes or not -- they hurt Marowak defensively much as they really, really help offensively because many things in GSC are just barely bulky enough to take one max-Attack hit for 90%+ -- it's that Marowak is the one offensive pokemon that cannot be countered defensively. Marowak is only beaten by KOing it before it deals too much damage itself. Skarmory is a decent short-term response but it will tend to lose eventually because it's caught Resting or gets flinched by Rock Slide. Surprisingly, Marowak even tends to have more defensive utility than the other offensive options due to its Electric immunity. Machamp doesn't really switch into anything but Ttar, Ttar doesn't really switch into anything except Lax in limited quantities, Dragonite's Ground immunity isn't much use beyond some Steelix (plus Spikes) and it relies on Dynamicpunch to differentitate itself offensively. Marowak directly preys on Raikou, especially Roar ones that call Marowak in against it, and gets Lax eating Hidden Powers instead of Thunders.

I still decided to put Machamp the next rank down because the ubiquity of Lax + Cloyster makes it very easy to build entire teams that can't eat a Cross Chop and its counters are still very limited if not at least existent. Starmie can always be blown up by Cloyster if the opponent overrelies on it as their sole Machamp counter, though odd things like Nidoking and Heracross can sometimes also prove meddlesome depending on game state. Machamp doesn't have much defensive utility itself though, it's pretty reliant on free/double switches and Leftovers to mitigate Spikes.

Skarmory, much as I malign "do-nothing" purely defensive mons, is at least the best defensive mon against the borderline broken undisputed #1 mon in the game. More importantly, it's not a complete dud offensively due to the threat from Curse. Steelix is pretty similar, but with a better Electric matchup at the cost of some others (dog to Marowak and Machamp, doesn't wall EQ Lax, etc.). Suicune only gets in the way of less threatening mixed attackers, such as Ttar and Nidoking, and has zero sweeping potential. Umbreon is not a threatening booster and the counter overlaps with its own actually threatening strategy, Trap Passing. And so on. Also, Ground immunity is cool for all you Spikes-obsessed nutters. I like Skarm + Zapdos teams for that reason.

Further on Steelix, sometimes based on matchup it's just a fucking rock that the opponent has no real response to. Even Skarmory doesn't tend to get that kind of matchup due to the omnipresence of Zapdos and Raikou. I do wish it was more threatening offensively, but then oftentimes I'll do something like use Rhydon in the same role and then immediately regret it in battle when it ends up eating some random fucking Toxic and be reminded as to why Steelix is more popular.

I only rated Charizard because it was required. I don't believe it's viable. Drum is difficult to set up and even when you do, you don't actually OHKO much from full HP and don't outspeed enough. Charizard has nothing else to offer over Moltres, which itself is terrible but the difference in power on STAB moves is notable enough to matter if you're actually considering using them.

I actually had the lowest Golem ranking at #30, not choolio, but if I wasn't overcompensating for the aggregate I'd put it somewhere in the 25-27 range.

I think Blissey is a better cleric than Miltank because many teams with Miltank tend to suffer severe matchup problems with the Electrics. Miltank itself can't fight them and additionally tends to be paired with more offensive Lax sets, Ttar over a Ground-type, etc. Perhaps more importantly, Blissey has a metric fuckton of versatility that is sorely underexplored due to low usage rate and being pigeonholed into a cleric role. Miltank can't do this.
 
Wait mr e if you dont find spikes essential to almost every team how is my #2 pick cloy “respectable”?
 

Mr.E

unban me from Discord
is a Two-Time Past SPL Champion
Raikou is simply worse than Zapdos. Cloyster is so radically different that it's a difficult comparison to make and I can respect someone who is willing to stand behind the notion that Cloyster's unique qualities (primarily Spikes and Explosion) give it the edge.
 

Users Who Are Viewing This Thread (Users: 1, Guests: 0)

Top