EPTI don't see why everyone is complaining so hard. From what I see there is nothing over a 5% deviation, which is pretty good, i'm sorry your race isn't on the uprise
[Aug] TLPD Race Winrate Graphs - Page 10
| Forum Index > SC2 General |
|
kodas
United States418 Posts
I don't see why everyone is complaining so hard. From what I see there is nothing over a 5% deviation, which is pretty good, i'm sorry your race isn't on the uprise | ||
|
Dbla08
United States211 Posts
| ||
|
Ihpares
United States40 Posts
Based on the sample size for August there's about a 2% Margin of Error. Basically, we can take it to mean that if the sample size were larger for the month of August, Terran could actually be around 50.4%, and Protoss around 48.9%. That's relatively balanced. That said, we haven't seen the Korean graphs. The win rates may not be so forgiving in Korea. | ||
|
Elefanto
Switzerland3584 Posts
On September 06 2011 10:20 Grumbels wrote: It's probably to be expected terrans will score better at the very highest level. How many really good terrans are there compared to zerg or protoss? I would say at least two times as much and honestly some of that must have to do with the popularity of the race right at release (when they were strongest and a lot of people were starting to play with the goal of becoming competitive) That's stupid, especially for Korea. People pick the race that's perceived to be the strongest with the most potential. And that has absolutely nothing to do with the overpopulation of Terrans. | ||
|
Tatari
United States1179 Posts
 ... If only that were true... T_T | ||
|
Cassel_Castle
United States820 Posts
http://www.teamliquid.net/tlpd/korean/leagues/118_Shinhan07_Proleague_R1 http://www.teamliquid.net/tlpd/korean/leagues/133_Shinhan07_Proleague_R2 http://www.teamliquid.net/tlpd/korean/leagues/147_Shinhan08_Proleague http://www.teamliquid.net/tlpd/korean/leagues/157_Shinhan08-09_Proleague http://www.teamliquid.net/tlpd/korean/leagues/274_Shinhan09-10_Proleague http://www.teamliquid.net/tlpd/korean/leagues/710_Shinhan10-11_Proleague SC2 has been balanced since April 2011 or longer. | ||
|
Amui
Canada10567 Posts
On September 06 2011 11:11 Cassel_Castle wrote: You'll never get 50% winrates for each race. 53-55% is perfectly normal for a balanced game, just look at BW. http://www.teamliquid.net/tlpd/korean/leagues/118_Shinhan07_Proleague_R1 http://www.teamliquid.net/tlpd/korean/leagues/133_Shinhan07_Proleague_R2 http://www.teamliquid.net/tlpd/korean/leagues/147_Shinhan08_Proleague http://www.teamliquid.net/tlpd/korean/leagues/157_Shinhan08-09_Proleague http://www.teamliquid.net/tlpd/korean/leagues/274_Shinhan09-10_Proleague http://www.teamliquid.net/tlpd/korean/leagues/710_Shinhan10-11_Proleague SC2 has been balanced since April 2011 or longer. Each one you took has less than 100 of each matchup. That's not really enough to judge because 1-2 games skews it heavily one way or another. From what I can see in those BW stats, each race is slightly favoured in one of their matchups as a trend. This isn't the case for the SC2 statistics. Terran is favoured in ALL matchups, and has been for months now. Protoss in particular is having a hard time of it. | ||
|
Shankapotamus
United States428 Posts
| ||
|
Not_That
287 Posts
| ||
|
Fig
United States1324 Posts
And don't say that you can't assume equal skill since there are so many "talented terrans." This wouldn't make any sense for there to be randomly twice as many skilled terran players compared to the other races. Therefore if they really are more talented then the only other explanation is that there are more people picking up terran in the first place, which isn't true based on stats about how many people play each race. It could be that the people who want to play competitively are the ones choosing terran more. But that would mean they thought it gave them the best chance to win, which if most of the up-and-comers do, then it is also bad for the game, because even they perceive it as the strongest race. | ||
|
Snorkle
United States1648 Posts
maybe we are? in that case I apologize and the only thing i edited in my post was a semicolon for clarity. When he says, 100 energy drained per emp, to me that sounds like 1 emp will only drain a total of 100 energy no matter how many units it hits. I wanted to clarify that it drains 100 energy from every unit it hits, so 1 EMP could drain upwords of 600 energy (per emp). Its just semantics if we are in agreement and this is derailment anyway so on topic: The international PvX winrates are close enough, the korean ones are a little bit far out of the margin of error. People say the sample size is too small but it is universally agreed that the concentration of good players and overall understanding of the game is much higher in korea, which gives added weight to their numbers. | ||
|
Ex
8 Posts
On September 06 2011 10:44 Ihpares wrote: http://en.wikipedia.org/wiki/Margin_of_error Based on the sample size for August there's about a 2% Margin of Error. Basically, we can take it to mean that if the sample size were larger for the month of August, Terran could actually be around 50.4%, and Protoss around 48.9%. That's relatively balanced. That said, we haven't seen the Korean graphs. The win rates may not be so forgiving in Korea. Or it could be 54.4 and 46.9 based on that logic. You can't just add and subtract wherever you want. If you use 2% as your confidence interval though, then all you can say is that you are confident that T wins 'more than it should' and vice versa for P. Besides, margin of error is only relevant when you are taking a sample from a larger population. In this case I would say that the entire population has been sampled (or close), namely official tournament games. So there is no margin of error. This of course raises questions about what you want your population to be, but I think what has been used is a fair trade off regarding play level and having a decent number of games. You definitely wouldn't want to include ladder stats of any level for example, even if it was possible. Talk of 'anything under 5% is balanced' is just ridiculous. Fluctuations in the rate are going to happen as strategies develop and this causes movements from month to month. However to suggest a long term average of 47.5 to 52.5 in a matchup is close enough to balanced if we are looking at a stable game is just wrong. It indicates a significant imbalance. At the moment though, I would say the development of strategies and patch changes mean its impossible to draw any real conclusions. All you can say is that T has without a doubt had it good the entire time since release with Z and P struggling at different points. I do think overall gameplay has started to settle down though. The level of play seen and builds executed are no longer all over the place. | ||
|
ieatfries
Canada44 Posts
 i question myself for playing this race | ||
|
Cone
United States5 Posts
On September 06 2011 10:15 Gegenschein wrote: People should learn not to put green and red in a graph, especially when there are only 3 colors involved. They'd make us color blind folks happy. Yeah, I agree with this. Especially because it was so clear during the earlier months and also because red-green color blindness is actually not all that uncommon (I think its almost like 1 in every 10 males is red-green colorblind). It used to be yellow for P, purple for Z and red for T and it was very vibrant and easily distinguishable, I think the colors actually suited each race better as well. I guess it wasn't traditional enough as far as the TL color icons went so they switched it to match those. | ||
|
Not_That
287 Posts
| ||
|
Gamegene
United States8308 Posts
On September 06 2011 12:55 Not_That wrote: He initially used different colors, until Day[9] remarked that he should follow the race color convention that has been used on TL since SC1. And honestly who wouldn't? | ||
|
mburke05
United States130 Posts
On September 06 2011 10:54 Elefanto wrote: That's stupid, especially for Korea. People pick the race that's perceived to be the strongest with the most potential. And that has absolutely nothing to do with the overpopulation of Terrans. actually he made alot of good points, and you just said "HEY NO ITS NOT THATS STUPID!! HEY!! STUPID!!" | ||
|
Not_That
287 Posts
| ||
|
Nothingtosay
United States875 Posts
| ||
|
Brett
Australia3822 Posts
The Korean stats are relevant. We've all looked at Korean Pro-BW stats for years as indicative of the general balance (with slight rock/paper/scissors features; T > Z > P > T) and they're a small sample size too. We've also used small numbers of matches to create map statistics to determine the balance of the specific map in BW too. Again, it's not really questioned, and maps that have shown win rates much > 70 odd % have generally found themselves quickly of of use... They (korean pros) are the pinnacle of the game and large variance in win % in their game pools is concerning. | ||
| ||
