EPTTop Tier Korean ZvT and TvZ TLPD statistics - Page 8
| Forum Index > SC2 General |
|
MostDifferent
Norway124 Posts
| ||
|
FrostedMiniWheats
United States30730 Posts
On March 16 2012 15:03 MostDifferent wrote: Lets talk about balance, based on stats that take games from old old patches into account!! balance? I was under the impression that this was a thread disguised to dick ride DRG like a jackhammer...my mistake | ||
|
Fiend13
Germany140 Posts
On March 16 2012 12:18 ETisME wrote: I am sorry but you got the stats wrong. The over 20 requirement is for cluster analysis, something that you aren't doing because you are not trying to make any clusters out from the data set. The over 20 thing you talked about is just for normal hypothesis testing, which you aren't doing. You need to calculate out the optimal minimal sample size based upon your confidence interval etc in short, you need to calculate out a sample size that truely represent the population. Merely 50 games out of his entire ZvT history does not make sense This in particular made my eyes bleed. I agree completely with you Sir or Madam. | ||
|
lazyitachi
1043 Posts
Random sampling = How representative of sample of demography studied. Hence since you are not studying the demography's average win rate this is irrelevant. You are looking at the TOP PLAYERS WIN RATE ACCORDING TO ELO hence not representativeness of random sampling to demography. Hypothesis testing: Given the probability of DRG's ZVT is (2/3) with 36 sample. Given the probability of MMA's TVZ is (52/69) with 69 sample. Null hypothesis: MMA wr > DRG wr at 5% probability of Type I error (p1 − p2) ± z * sqrt ((p1 q1)/n1 + (p2 q2)/n2 ) where n = no of games, p = win, q = loss, 1 = MMA, 2 = DRG, z = Standard score therefore substituting in ( 75.36% - 66.67%) ± z * sqrt ( 75.36% * 24.64% / 69 + 66.67% * 33.33% / 36) At 90% confidence level, the probability of making type I error is 20.8% At 95% confidence level, the probability of making type I error is 24% Hence your comparison is not statistically significant if you only tolerate 5% error You can group the data for top tier Ts and Zs for comparison. I doubt you have enough data for any statistical significance at individual player level. Credibility theory states the probability of each individual win rate being correct is = 2 * z ( k * sqrt (n)) - 1 where z = Standard Score, k = probability Type I, n = number of games We assume 10% Type I error i.e. k = 5% (divide by 2 because two-tailed test) DRG: = 2 * z( 5% * sqrt(36)) - 1 = 23.6% MMA: = 32.2% This means that the probability of DRG and MMA's win rate being the expected win rate is only 24% and 32% assuming 10% Type I error i.e. NOT ENOUGH DATA. | ||
|
sluggaslamoo
Australia4494 Posts
On March 16 2012 13:05 Veldril wrote: One thing to consider DRG's stat is that he lost to MMA (80% win rate against Zerg) 4 times in the Blizzard Cup final and he performs quite badly during his early career in the GSL.And from what I checked, he only has negative win rate against MMA and Clide (in GSTL). So yeah, he's really good at ZvT. He's just play top tier players more. As for why Terran is good at ZvT, well I would say that there's pretty much huge gap in the ability between Zerg players of GSL Code S and the rest. For example, Taeja's high ZvT % comes from ESV Korean Weekly, where he wins against someone like Jookto or Seal or Hyun. Those three players are completely on the other level compare to DRG or Nestea. 64% is already very very good. I present you the top 5 stats of BW's pros (according to ELO): Flash: 71.49% Bisu: 65.19% Fantasy: 61.80% Jaedong: 68.13% Leta: 59.01% As you can see, most players hover around 60-68% (except God). 64% in one match up is not bad at all for DRG (considering he's over 68.75% win rate). Shouldn't it be 50 games TvZ considering that's how DRG's stats are calculated? Flash (TvZ) - 36 wins - 13 losses (73.47%) Jaedong (ZvT) - 31 wins - 15 losses (67.39%) Bisu (PvZ) 46 wins - 11 losses (80.70%) <<< The only Protoss that understands PvZ Stork (PvZ) 28 wins - 22 losses (56.00%) Savior 2006 (ZvT) (1/12/2005 > 1/12/2006) - 30 wins - 8 losses (78.95%) [If he didn't cheat or get bored of SC he might have made quite an impact in SC2] | ||
|
itsjuspeter
United States668 Posts
| ||
|
guN-viCe
United States687 Posts
1) For a long time T has been considered slightly imbalanced in Korea due to a higher "skill ceiling". This makes sense because pretty much every T unit requires micro to become cost effective and supply effective(barring maybe Thors). Don't get me wrong, the other races require insane micro too, but not nearly to the same degree. 2) MMA is an outlier. If Flash played zerg would he be as dominate? Is Flash just intrinsically better? Same thing with MMA. Maybe he is just that good. Maybe DRG is amazing, but his true skill is comparable to a lesser skilled terran like MKP. If DRG switched to T, would he be as good as MMA? 3) The data included extends through multiple patches. This makes it harder to draw firm conclusions. | ||
|
Fubi
2228 Posts
On March 16 2012 11:14 neoghaleon55 wrote: Statistics is very dependent on standard deviations which accounts for your confidence interval. I like how the OP is now wiki-linking to standard deviation, then leaves out standard deviation from his analysis all together. That 5% difference could completely be within the margin of error of your statistics that you presented, and therefore making your results insignificant in statistical terms. | ||
|
-Exalt-
United States972 Posts
It's also funny that the insane amount of Terrans (and zergs) complaining about Protoss recently seem to ignore these stats and never stop complaining / bming on ladder about Protoss. ohh sc2 community, y so ignorant of this thing called "metagame". give it a month and P will be back to the weakest race. | ||
|
FidoDido
United States1292 Posts
you're saying the game is imbalanced because the top 5 terrans are getting 65% avg win rate in TvZ (excluding MMA because he understands the matchup too well) what about how all the zergs you listed all have over 65% WR in ZvP? You should argue that they should fix ZvP so it isn't op for zergs before you complain about Zergs being weak in ZvT. | ||
|
yourteam
United States40 Posts
N > 20 w000t job's done i can publish now LMFAO........... | ||
|
Chytilova
United States790 Posts
| ||
|
Logros
Netherlands9913 Posts
| ||
|
Fubi
2228 Posts
| ||
|
flowSthead
1065 Posts
On March 16 2012 13:04 neoghaleon55 wrote: Happy like 27th place in TLPD...I'm only taking into account top 20 of the korean TLPD... there are not even that many zergs to talk about the farther you go out....and they're not top tier. Great post by Mr. Nefarious. I featured your comment in the OP. This makes no sense to me. If anything, including Happy makes your point even more, and the idea that because a player isn't in the top 20 of the Korean ELO they aren't top tier is ridiculous. Squirtle, Puzzle, Crank, Inca, and Extreme are all higher than Parting at the moment, are you really suggesting that they are higher tier than Parting? Probably not. TLPD is at least somewhat dependent on past results, so it might not be completely accurate as to current skill. Plus, all of those players, including the ones not in the top 20, are playing each other. Happy has been playing Code S and Code A level zergs and he has an 83% winrate. Shouldn't that be recognized? | ||
|
neoghaleon55
United States7435 Posts
On March 16 2012 16:47 flowSthead wrote: This makes no sense to me. If anything, including Happy makes your point even more, and the idea that because a player isn't in the top 20 of the Korean ELO they aren't top tier is ridiculous. Squirtle, Puzzle, Crank, Inca, and Extreme are all higher than Parting at the moment, are you really suggesting that they are higher tier than Parting? Probably not. TLPD is at least somewhat dependent on past results, so it might not be completely accurate as to current skill. Plus, all of those players, including the ones not in the top 20, are playing each other. Happy has been playing Code S and Code A level zergs and he has an 83% winrate. Shouldn't that be recognized? It's hard to quantify skills without numbers. You can say happy's better than everyone, but the numbers do not show it. Happy may be a good ZvT-er but his overall ELO stat does not measure up. In the end you will have to rely on data to argue anything. | ||
|
neoghaleon55
United States7435 Posts
On March 16 2012 15:31 Fubi wrote: I like how the OP is now wiki-linking to standard deviation, then leaves out standard deviation from his analysis all together. That 5% difference could completely be within the margin of error of your statistics that you presented, and therefore making your results insignificant in statistical terms. It's definitely more than 5%...I just rounded down. Maybe I should write this again, since people are not even reading the numbers. | ||
|
flowSthead
1065 Posts
On March 16 2012 16:55 neoghaleon55 wrote: It's hard to quantify skills without numbers. You can say happy's better than everyone, but the numbers do not show it. Happy may be a good ZvT-er but his overall ELO stat does not measure up. In the end you will have to rely on data to argue anything. But his overall ELO is a measure of his other matchups as well, not TvZ. And since you are only looking at the TvZ matchup, that should be the only relevant thing. Who cares how he does against Protoss and Terrans when you are interested in how he does against Zergs? | ||
|
bastqq
Ukraine12 Posts
All other zergs has not so big winrate (relative to the terrans) becouse in relise sc2 was not best balanse and map pool. And all the way around with the Terrans | ||
|
sc14s
United States5052 Posts
On March 16 2012 11:20 Liquid`Jinro wrote: How the fuck is 64% not good? it's not good compared to say mma's win rate.. | ||
| ||
