Alright, November stats for Shoddy are up, let's discuss

Dragontamer said:
I'd say Smogon should look for long-term trends when making the OU list.
Click to expand...

Exactly. A weighted list containing the past 12 months (not right now, obviously) would be the best way to go. I commented it in a previous thread (maybe last month's shoddy statistics thread?), and with the proper constants, I think it would be a neat way of measuring OU.


Oh, and by weighted I mean all 12 months count, but the older they are, the less they count.
 
Colonel M said:
Code: 
   1. Gyarados (18.96 %)
   2. Hippowdon (16.93 %)
   [B][I]3. Salamence (14.91 %)[/I][/B] <----- I can sort of understand this, but I usually find Salamence to be a better trump card than a starter (but Intimidate is nifty to have).
   [B]4. Gengar (14.46 %)[/B] <----- Scarf, okay.  Anything else... blech.  As such, shouldn't Gengar be afraid of [B]NUMBER 5: TYRANITAR!?![/B]
   5. Tyranitar (13.69 %)
   6. Bronzong (11.6 %)
   7. Weavile (11.42 %)
   8. Swampert (11.41 %)
   [B]9. Infernape (9.66 %)[/B]
  10. Abomasnow (9.63 %)
  11. Yanmega (9.3 %)
  12. Breloom (9.16 %) <----- I can understand this if it's a Double-Powder lead
  13. Forretress (8.69 %)
  14. Ninjask (8.57 %)
  15. Roserade (6.82 %)
  16. Crobat (6.54 %) <----- Ah, seems like Crobat is slowly climbing into glory
  17. Metagross (6.25 %)
  18. Jolteon (6.23 %)
  19. Ambipom (6.23 %) <---- 95% mark
  20. Staraptor (4.86 %)
  21. Zapdos (4.68 %)
  22. Aerodactyl (4.61 %)
  23. Azelf (4.43 %)
  [B]24. Garchomp (4.19 %)[/B] <----- Ok... Now I'm REALLY LOST!
  25. Alakazam (4 %)
  [B]26. Heracross (3.99 %)[/B] <----- Fallen quite a bit, have we?
  27. Togekiss (3.85 %)
  28. Skarmory (3.78 %)
  29. Smeargle (3.44 %)
  30. Machamp (3.19 %)
  [B]31. Starmie (3.1 %)[/B] <----- ...No
  32. Scizor (2.95 %)
  33. Porygonz (2.88 %)
  34. Typhlosion (2.79 %)
  35. Lucario (2.54 %)
  36. Spiritomb (2.41 %)
  37. Donphan (2.39 %)
  38. Kingdra (2.28 %)
  39. Medicham (2.28 %)
  40. Heatran (2.27 %)
  41. Tauros (2.12 %)
  42. Sceptile (2.1 %)
  43. Jirachi (1.98 %)
  44. Umbreon (1.92 %)
  45. Persian (1.75 %)
  46. Shuckle (1.73 %)
  47. Flygon (1.66 %)
  48. Gliscor (1.57 %)
  49. Tentacruel (1.57 %)
  50. Froslass (1.56 %)
  51. Floatzel (1.52 %)
  52. Gallade (1.46 %)
  53. Venusaur (1.42 %)
  54. Jumpluff (1.32 %)
  55. Nidoqueen (1.29 %)
  56. Dusknoir (1.27 %)
  57. Butterfree (1.21 %)
  58. Magmortar (1.19 %)
  59. Mamoswine (1.18 %)
  60. Hariyama (1.13 %)
  61. Arcanine (1.11 %)
  62. Hitmonlee (1.08 %)
  63. Charizard (1.04 %)
  64. Nidoking (1.03 %)
  65. Cresselia (1.02 %)
  66. Slaking (1.02 %)
  67. Dragonite (1.02 %)
  68. Rhyperior (1.02 %)
The ones in bold are more WTF ones, and I sort of added some input. I think I'm going to attempt to look over the starters of each month and use them to change the guide from time to time.
Click to expand...

Garchomp Maybe due to the gyarados leads he is maybe forced to be used less as a lead. D.meteor has a chance of no KO'in a bulky gyarados and outrage is even worse because of intimidate. So they are probably keeping him as a trump

Gengar he still ohko t-tar with focus blast (even with a wide lens)

Starmie not sure if you meant the no in a positive or negative way
 
Hey Colin... if you're still reading this thread, can we get a feature added to Shoddy to prevent you from battling the same person in back-to-back matches. I'd rather play someone much better or worse than myself than play the same opponent 3 and 4 times in a row.

Also, Froslass is the hotness on a hail team. With Snow Cloak on both her and Mamoswine, hax tend to happen very frequently. So while her stats and "normal" efficiency isn't that high, when she's used in the right environment, Froslass holds her own very well with OU pokes. I think she could at least be safely called BL considering her usage and abilities.
 
Why has the 95% mark been chosen? Are we assuming the usage statistics are a normal distribution?

Also, I can't believe people still argue Abomasnow is not ou...
 
The 95% mark hasn't been confirmed yet, but it's kinda the mark people are talking about the most.

Dragontamer said:
Anyway, the "total" is the total number of usages, as determined by blah.pl (that is, 34825 (Blissey) + 28071(garchomp) + etc. etc. etc. Blah2.pl generates the text. I change this line $tmp = $2 / 83436.0 * 10000; to whatever the total was. In November, the total happened to be 83436 (as blah.pl calculated).

Yeah, I need better names for the programs, but they're like, 5 lines long >_>
Click to expand...
I'm not getting 83436. I'm getting 763551 when I add up the total.
 
Yeah, Abomasnow is obviously OU, even if he shouldn't be. It's based on usage, not strength.

Also, I wonder if Shoddy can get the most popular movesets for each Pokemon...
 
SHUCKLE MAN said:
Yeah, Abomasnow is obviously OU, even if he shouldn't be. It's based on usage, not strength.

Also, I wonder if Shoddy can get the most popular movesets for each Pokemon...
Click to expand...

Judging by what Colin said about how the data is collected (going over battle logs) I'd say it's *possible*, but...

I doubt it'd tell us anything we didn't already know.
 
Though Abomasnow is the 10th most popular lead, which is certainly very high, note that it is almost used exclusively as a leading Pokemon, as the fact that it's in the 40's (IIRC) in the normal usage list evidences. Additionally, if Shoddy counted even the Pokemon that aren't sent out, the usage would likely sink further, because Abomasnow, as said before, is used mainly as a lead.
 
The 763551 is the sum I calculated as well. And 158518 for the leads. The 80k sum is smaller than the sum of the top 6 pokemon appearances on the list.

Also, since each lead team should have 5 other pokes, the appearances total about 80% of the maximum. So there is a fraction of the unknown pokemon that are never seen.
 
Hmm. Polytree adding the number of leads gives me an idea. We should use the total number of leads (158518) as the total number of teams that were used.
 
Okay, I decided to think about this a bit more.

What I did is as follows. I first added up all the usages together, getting 763551 usages, and then found the percentage usage of each Pokemon. Then I made these as a cumulative frequency percentage, and tabulated them:

Code: 
Blissey    4.56
Garchomp   8.24
Gengar     11.31
Tyranitar  14.01
Gyarados   16.61
Salamence  19.1
Infernape  21.57
Skarmory   23.87
Cresselia  26.08
Bronzong   28.25
Weavile    30.39
Forretress 32.42
Metagross  34.36
Swampert   36.27
Starmie    38.17
Heatran    40.01
Gliscor    41.71
Heracross  43.33
Breloom    44.93
Dusknoir   46.4
Lucario    47.8
Hippowdon  49.2
Electivire 50.51
Togekiss   51.74
Donphan    52.97
Porygonz   54.18
Azelf      55.32
Tentacruel 56.45
Yanmega    57.51
Vaporeon   58.5
Scizor     59.48
Jirachi    60.42
Magnezone  61.35
Spiritomb  62.25
Mamoswine  63.09
Milotic    63.9
Dragonite  64.7
Snorlax    65.49
Weezing    66.27
Machamp    67.01
Ninjask    67.72
Jolteon    68.42
Zapdos     69.11
Roserade   69.78
Dugtrio    70.45
Abomasnow  71.1
Alakazam   71.75
Celebi     72.38
Umbreon    73
Suicune    73.61
Gallade    74.19
Froslass   74.75 <--- SUGGESTED CUTOFF POINT FOR OU
Ambipom    75.3
Cradily    75.81
Crobat     76.31
Claydol    76.81
Rhyperior  77.3
Sceptile   77.78
Aerodactyl 78.26
Tangrowth  78.73
Kingdra    79.2
Slowbro    79.65
Staraptor  80.11
Empoleon   80.56
Charizard  81
Blastoise  81.39
Arcanine   81.79
Mismagius  82.18
Clefable   82.55
Houndoom   82.93
Azumarill  83.28
Smeargle   83.64
Miltank    83.99
Medicham   84.33
Typhlosion 84.66
Ludicolo   84.96
Poliwrath  85.25
Gardevoir  85.53
Flygon     85.81
Hitmontop  86.08
Raikou     86.35
Hariyama   86.62
Blaziken   86.87
Shuckle    87.12
Feraligatr 87.37
Steelix    87.62
Walrein    87.87
Toxicroak  88.11
Venusaur   88.35
Rotom      88.59
Drapion    88.82
Cloyster   89.06
Magmortar  89.29
Ninetales  89.51
Rampardos  89.73
Marowak    89.94
Jumpluff   90.15
Hitmonchan 90.36
Porygon2   90.56
Swellow    90.77
The above are only the top 100 Pokemon.

I then suggest that the cut-off point for OU are those Pokemon whose cumulative percentage usage is 75% of the total usage or less. This means that if a Pokemon is used, there is a 75% chance that it's one of those 52 and only a 25% chance that it is one of the other 441. This would make OU have 52 Pokemon.

Alternatively, the weighted Shoddy battle list is used instead. I would actually prefer to use the weighted list instead of the usage list, since it prefers winning Pokemon instead of just the used ones. Here is the cumulative percentage list using the weighted list:

Code: 
Blissey    4.79
Garchomp   8.58
Gengar     11.59
Tyranitar  14.52
Gyarados   17.15
Salamence  19.7
Infernape  22.19
Cresselia  24.59
Skarmory   26.94
Bronzong   29.23
Weavile    31.28
Metagross  33.3
Heatran    35.29
Swampert   37.29
Forretress 39.27
Starmie    41.24
Gliscor    43.07
Breloom    44.76
Heracross  46.44
Hippowdon  47.9
Dusknoir   49.36
Lucario    50.79
Donphan    52.05
Togekiss   53.25
Azelf      54.45
Tentacruel 55.64
Electivire 56.76
Porygonz   57.87
Yanmega    58.92
Vaporeon   59.93
Jirachi    60.94
Scizor     61.93
Magnezone  62.91
Spiritomb  63.84
Mamoswine  64.7
Milotic    65.5
Dragonite  66.29
Machamp    67.07
Zapdos     67.82
Weezing    68.58
Snorlax    69.32
Roserade   70.06
Dugtrio    70.75
Celebi     71.43
Jolteon    72.1
Abomasnow  72.74
Suicune    73.38
Ninjask    73.96
Alakazam   74.53 <--- SUGGESTED CUT-OFF POINT
Gallade    75.09
Umbreon    75.63
Froslass   76.17
Cradily    76.69
Crobat     77.21
Ambipom    77.72
Claydol    78.22
Tangrowth  78.7
Rhyperior  79.17
Staraptor  79.63
Empoleon   80.08
Sceptile   80.53
Slowbro    80.97
Kingdra    81.41
Aerodactyl 81.83
Charizard  82.23
Mismagius  82.62
Clefable   83
Blastoise  83.37
Miltank    83.74
Houndoom   84.1
Arcanine   84.45
Azumarill  84.79
Smeargle   85.12
Medicham   85.44
Poliwrath  85.76
Typhlosion 86.07
Ludicolo   86.37
Hitmontop  86.64
Raikou     86.92
Gardevoir  87.18
Hariyama   87.44
Flygon     87.69
Shuckle    87.95
Rotom      88.2
Walrein    88.45
Toxicroak  88.69
Feraligatr 88.92
Blaziken   89.15
Steelix    89.38
Cloyster   89.6
Venusaur   89.82
Porygon2   90.03
Drapion    90.23
Jumpluff   90.44
Hitmonchan 90.64
Ninetales  90.85
Magmortar  91.04
Marowak    91.24
Rampardos  91.42
Swellow    91.61
If the weighted list is adopted, there would only be 49 OU Pokemon (Umbreon, Gallade and Froslass wouldn't make the cut).

I dunno what you think of this.
 
Very nice statistics, X-Act. I, too, recommend using the weighted list, because of how oblivious and manipulable the general usage list is. In effect, the general list ends up recording more of which Pokemon are used by people who play more games than which Pokemon are truly effective. To clarify, a poor player who plays hundreds of games would have more representation in the general list than decent players who have a good winning percentage but don't spend as much time playing.
 
Unless my eyes are decieving me, Hitmonchan and Hitmontop are BL (or UU if you will), yet Hitmonlee doesn't even appear here. How odd...

Hm, this means a lot of oddball Pokemon are BL / UU:

- Tangrowth, which has sadly lost a bit of popularity
- Slowbro, who was at one point OU IIRC.

Those two are ok in the BL limbo. These two, however, a little bit questionable:

- Raikou, which is seriously... what... the... hell!?! Even Suicune barely made the cut, and some argue Raikou is better.
- Rhyperior, which has Solid Rock fixed now on Shoddy.

The list for now would be fine. I'd still argue how a couple on the "cut-off" list Pokemon as well, as Alakazam being OU strikes me as the wierdest thing. Before I rarely saw an Alakazam. >_>; Still, BL is a faux tier.

For the above poster that questioned me on Starmie being a lead, I meant it in a negative way. Starmie should never start in the battlefield at all.
 
IMO, use the Cutoff percentage as a baseline, then certain "arguable" pokes BELOW the cutoff point (Raikou is a perfect example), and either shift the boundary between OU/BL to that point, or merely move the argued pokemon up to OU.

Or, remove BL all together, and make it purely OU -> UU (Please don't kill me >_>)
 
i'm not suprised to see gyarados, infernape, gengar, weavile, skarmory and salamance as leads. i am suprised to see abomasnow and bronzong. i think that i have only seen one abomasnow when battling. but i have never seen a bronzong as a lead.
 
bronzong is a good lead as it can hypnosis AND stealth rock. it can also run 2 attacking moves that can do a decent amount of damage with bronzong's attack.

i used to run a hypnosis stealth rock charge beam HP ice set to coutnre gyara leads. it pwned against noobs..

abomasnow starts off hail first turn, and it is a pretty damn powerful choice scarfer too. it has an average base attack, but has good stabs and had very, very good coverage,
 
well, i have had a lot of battles, but have never seen them as leads. goes to show that most people use the typical leads like skarmory, gyarados, weavile, etc. all OU.
 
Zenigame said:
What happened to Rhyperior? Why is Shuckle beating it? Abra beats Rhydon? Oh well, I don't play shoddy.
Click to expand...

Rhyperior is a retarded lead. Only a retarded nut would use Rhyperior as a lead.
Ok lawl, seriously, I use Infernape as a lead. And call me a weirdo, but I use specs on it. 104 base spA+STAB+140 base power+specs=boom. Close to the power of Specsmence Draco Meteor. I use HP Electric/Overheat/Grassknot/Close Combat and it gets so many of the common leads screwed over, like Swampert, Hippo, etc. I once rofl'ed when I ohko'ed an unsuspecting Gyara lead with HP Electric. Not to mention Overheat did 826% damage to a Magnezone.
 
It is technically not "the OU Pokemon are used by 75% of people" - it is the most popular pokemon that fill 75% of the slots. I was trying to figure out probabilities (as in, OU pokemon would appear 1 in x number of matches), but I gave up for today after the probabilities I used didn't sum to one.

However, using those top 49, here are some random facts.

- 14 are immune to Ground, 9 to Electric
- Fire is SE on 17 of 49, the most of any attack type
- Ice(14) is next, with four 4x SE.
- Least resisted? Dragon (9)
- Others? Flying (10), Water (11), Psychic (8, 3 immune)

- 28 of 49 have Steel (9), Psychic(8), Ground(7) or Flying(9) as a type
- 37 have an immunity from at least one type
- Tyranitar is the only Rock type.
 
You're right. It is not "OU are used by 75% of people". I typed that when it was 1am for me.

It is "If a Pokemon is used, there is a 75% chance that it's one of those 49 or 52 (whichever list you use), and only a 25% chance that it is one of the other 444 or 441".

If that's not a definition of OU, I don't know what is.

Feel free to tweak with the percentage as you see fit. I think 75% is a good all-round number, but maybe 80%, 70%, 66%, 60%, etc. might be better numbers. Suggest as you prefer.

If, say, 80% is used as the cutoff point, then there will be 10 more Pokemon that will be considered as OU in both the unweighted and the weighted lists.
 
Bumping this for a few other charts. :)

Since the weighted rank is the sum of all the ranks of the players who used that Pokemon, the average ranking of the player that used that Pokemon can be worked out by dividing this number by the number of times the same Pokemon was used.

This can be used to find which are the Pokemon used by the top-ranked players and those used by the bottom-ranked players. To this end, I averaged the ranks and found their standard deviation. I only used the Pokemon that had 30 or more usages by the way (so that it was used at least once per day). Then, I listed those Pokemon that are used by players ranked higher than (mean + stdev) to be the Pokemon used by the top-ranked players. Those Pokemon that are used by players ranked below (mean - stdev) are those used by the bottom-ranked players.

Without further ado, here are the lists, together with their usage position:

Code: 
POKEMON USED BY TOP-RANKED PLAYERS IN NOV-07:

1.   Blissey
4.   Tyranitar
9.   Cresselia
10.  Bronzong
16.  Heatran
17.  Gliscor
19.  Breloom
22.  Hippowdon
27.  Azelf
28.  Tentacruel
32.  Jirachi
33.  Magnezone
40.  Machamp
43.  Zapdos
44.  Roserade
48.  Celebi
50.  Suicune
73.  Miltank
77.  Poliwrath
90.  Rotom
117. Mesprit
119. Muk
128. Moltres
134. Probopass
158. Golem
250. Murkrow
Code: 
POKEMON USED BY BOTTOM-RANKED PLAYERS IN NOV-07:

102. Espeon
161. Aggron
165. Luxray
172. Flareon
176. Altaria
184. Bastiodon
187. Arbok
193. Xatu
194. Rattata
197. Dewgong
201. Grumpig
209. Vespiquen
211. Chimecho
213. Kecleon
214. Seviper
215. Crawdaunt
216. Farfetch'd
220. Squirtle
223. Ditto
226. Pidgeot
230. Whiscash
233. Bulbasaur
235. Mawile
236. Castform
237. Charmander
239. Granbull
240. Sableye
242. Bellossom
244. Ledian
245. Cherrim
246. Shuppet
247. Exploud
251. Stantler
255. Wartortle
256. Vigoroth
259. Magmar
265. Fearow
 
You must log in or register to reply here.
Back
Top