EPT:/
July Winrates! - Page 9
| Forum Index > SC2 General |
|
nkr
Sweden5451 Posts
:/ | ||
|
dani`
Netherlands2402 Posts
On August 03 2013 19:25 nkr wrote: Stop using the word sample size when the data is the whole population that is being examined. It's a census :/ Considering Aligulac has between 1300 and 1700 games in July per matchup, it appears to me data in the OP is fact a sample of all available tournament games since it has only 400-500 games per matchup (Korea + international). | ||
|
Grovbolle
Denmark3813 Posts
On August 03 2013 19:33 dani` wrote: Considering Aligulac has between 1300 and 1700 games in July per matchup, it appears to me data in the OP is fact a sample of all available tournament games since it has only 400-500 games per matchup (Korea + international). OP contains: All Premier Tournaments All Major Tournaments All Direct Qualifiers to Premier Tournaments (this does not include Qualifiers for Qualifiers *only incl. later stages) All Premier Teamleagues (GSTL, ATC and PL) All Monthly Finals (Go4SC2 and Zotac Top 16, no weekly tournaments) Was listed in the reddit thread. | ||
|
keglu
Poland485 Posts
On August 03 2013 22:20 Grovbolle wrote: OP contains: All Premier Tournaments All Major Tournaments All Direct Qualifiers to Premier Tournaments (this does not include Qualifiers for Qualifiers *only incl. later stages) All Premier Teamleagues (GSTL, ATC and PL) All Monthly Finals (Go4SC2 and Zotac Top 16, no weekly tournaments) Was listed in the reddit thread. Thats still 1/3 of games so clearly not whole population Also WCG is listed as premier tournament, i dont see qualifiers to WCG listed. | ||
|
Grovbolle
Denmark3813 Posts
On August 03 2013 22:25 keglu wrote: Thats still 1/3 of games so clearly not whole population Also WCG is listed as premier tournament, i dont see qualifiers to WCG listed. Just answering this part of Dani's post: it appears to me data in the OP is fact a sample of all available tournament games | ||
|
nkr
Sweden5451 Posts
On August 03 2013 22:25 keglu wrote: Thats still 1/3 of games so clearly not whole population Also WCG is listed as premier tournament, i dont see qualifiers to WCG listed. It is because the winrates states for example "winrates korea". If you then use winrates using all the matches in korea then that's not a sample, that's a census. If you used all the matches in korea and then used it to represent the world wide winrate, then the matches in korea would be a sample, and the amount of matches the sample size. It would be a horrible way to take a sample, but a sample none the less. | ||
|
Apsilon
Germany7 Posts
| ||
|
thezanursic
5497 Posts
edit: OHH shit korea is fucked up | ||
|
xyzz
567 Posts
| ||
|
Markwerf
Netherlands3728 Posts
On August 03 2013 19:25 nkr wrote: Stop using the word sample size when the data is the whole population that is being examined. It's a census :/ it's not a census obviously because the data is only from a period of time and is used in a discussion about general balance.. Sample size is a fine word. | ||
|
Doublemint
Austria8704 Posts
On August 04 2013 06:50 xyzz wrote: Since Taeja won ASUS ROG, HOTS Terran now have 6 major tournament victories, Zerg 5, and Protoss 2. I doubt it can be a coincidence for years and years that Protoss simply isn't capable of competing in the long run for the premier trophies. If the race works for lesser players on ladder and in casual tournaments, but not for high-end professionals, then the skillcap and potential of the race simply isn't as high as the other two. How funny it would be if San came back from that  | ||
|
xyzz
567 Posts
How funny it would be if San came back from that That guy plays Protoss. The no comebacks race. | ||
|
nkr
Sweden5451 Posts
On August 04 2013 06:56 Markwerf wrote: it's not a census obviously because the data is only from a period of time and is used in a discussion about general balance.. Sample size is a fine word. but the data only describes that period of time, it's obviously a census | ||
|
Lumi
United States1616 Posts
I'm glad that none of you have anything to do with actually balancing this game. | ||
|
Markwerf
Netherlands3728 Posts
On August 04 2013 07:56 nkr wrote: but the data only describes that period of time, it's obviously a census doesn't matter as a small portion of all theoretical available data is used to say something about the general population paramater, ie balance. Balance is just the winrates from the theoretical set of infinite matches. Thus in that discussion this is definately not a census and the word sample size is absolutely fine. If you were to talk only about July and these specific set of games then sure this is a census but I think it's safe to assume this data is used as part of a discussion about general trends/balance. It's about context and in the way virtually everyone uses it here sample size is correct. The way you look at it every piece of data ever is a census because data is always complete for some restricted set. Either way the general point most make which is completely fair is this numbers are just a little piece and especially since they are not even independent (as lots of players played multiple games) you can't say much from this data. The OP does a fine job not trying to conclude anything from it but still these 'statistical' posts make me cringe most of the time as it's too easy for everyone to draw wrong conclusions. | ||
|
doggy
Germany306 Posts
On August 02 2013 14:02 TAMinator wrote: If zerg is gonna get buffed itll probably be the viper, ive seen 2 interviews where d kim has eluded that the viper would be their priority if zergs were having trouble The only problem i see with buffing viper is that it wont help muta ling bane in zvt at all, which is the most entertaining unit composition against bio mine to play and watch. Roach hydra is a lot more boring imo. However i cant see a other way (maybe infestor?) to buff zerg without ruining the balance for other matchups. | ||
|
Incognoto
France10239 Posts
On August 01 2013 15:26 blade55555 wrote: lol shows how much you pay attention to zerg. In zvt, I have tried swarmhosts a lot, and if you go swarmhosts drop play murders you. If you go ling/infestor you get crushed by bio/mine because infestors aren't as good as they used to be. Like you act like we zergs are just sticking to wol strats, no if that were true we would be doing ling/infestor into bl/corr/infestor. No zergs are experimenting and trying other stuff, but non muta play in zvt makes zvt a lot harder. zvp you see lots of zerg uses with vipers/swarmhosts/mutalisks/etc. Like I don't think you know what you are talkinga bout. Correct me if I'm wrong, but here's my analysis of ZvT. In early WoL, the go-to in ZvT was Marine/Med/Tank vs Ling/Bling/Muta, which was more or less balanced. However after Queen patch, Zergs started using infestors (or Infestor/Brood) and started to dominate the match up. After they nerfed the infestor in Hots, they also buffed Muta. But Mutalisks, even with their speed and HP regen, can't harass like it was possible in WoL due to widow mines. The problem being that widow mines do well against all three units of the Ling/Bling/Muta composition. Furthermore, Zerg's only true answer to drop harass are Mutalisks, but if they only serve a defensive purpose then that's a lot of resources that can't be used when Terran pushes (ie in a straight up fight). What are the Mutalisks target? In WoL they went for Tanks when facing Marine/Med/Tank, but in Bio/mine, both units seemingly counter the mutalisk. So ZvT is either Zerg being able to fight straight up on the ground with Roaches, Infestors, Ultras, Blings and can't deal with drops effectively since they have no "true" anti-air or Zerg deals with drop harass but can't fight straight up efficiently because Mutalisks are mostly dead weight in a straight up fight. That's the fundamental difference between Marine/Tank and Bio/Mine, I honestly think that the mine is what's causing problems in the match up, but again, I may be slightly biased here. This is the impression I have, not through my own games but watching professional play. | ||
|
FeyFey
Germany10114 Posts
On August 04 2013 06:50 xyzz wrote: Since Taeja won ASUS ROG, HOTS Terran now have 6 major tournament victories, Zerg 5, and Protoss 2. I doubt it can be a coincidence for years and years that Protoss simply isn't capable of competing in the long run for the premier trophies. If the race works for lesser players on ladder and in casual tournaments, but not for high-end professionals, then the skillcap and potential of the race simply isn't as high as the other two. Or Protoss is just easier to learn and harder to master then the other 2 races, since the skillcap in sc2 can never be reached by any mortal. The only skill cap is the physical limit of the people playing it, which is shifting heavily so balancing around it is not a good idea. The race itself has many "sneaky" units though and afterwards there is just a small amount of straight up units left, that do well even if scouted. It makes it easy to train against one style that every protoss uses in the lategame. Just like Terrans in TvP that don't have many options either. Most Protoss play behind on upgrades always, I am curious how the winrates would be if they wouldn't try it with 3/3 only. So I would go with alot of unused potential left, just like Zergs not using Burrow. Though burrow would need a whole new layer of thinking, while 3/3/3 Protoss units are just more badass and the only thing you have to remember is to get +1 shield before +3 armor. | ||
|
Big J
Austria16289 Posts
On August 03 2013 22:25 keglu wrote: Thats still 1/3 of games so clearly not whole population Also WCG is listed as premier tournament, i dont see qualifiers to WCG listed. well, aligulac contains all sorts of small tournaments, which screws with balance. As in those tournaments there are very many "Pro vs NonPro" or even just "Master vs Platinum" situations, in which the latter player will lose 99% of the time, regardless of balance. An example what these kind of things do: matchup: AvB is 60:40 balanced with 1000 games in which the players are evenly matched. Then there are 500games in which the winrates are 99% for and 500games in which the winrates are 1% for A. winrate of race A is therefore shown as (60*1000+99*500+1*500)/2000 = 55% instead of the true 60% balance. That's why stats with "lesser tournaments" will always be close to 50:50. As games in which one player is "outmatched" exist for both races and therefore the balance of those games is 50:50, the stats always tend to be closer to 50:50 as they should be from "Pro vs Pro" stats alone. Of course it remains to find out how big this influence really is. At least that's one approach to explain why aligulac (and other big statistics) never really has any outliers. | ||
Liquid`Ret
Netherlands4514 Posts
On August 04 2013 19:59 Big J wrote: well, aligulac contains all sorts of small tournaments, which screws with balance. As in those tournaments there are very many "Pro vs NonPro" or even just "Master vs Platinum" situations, in which the latter player will lose 99% of the time, regardless of balance. An example what these kind of things do: matchup: AvB is 60:40 balanced with 1000 games in which the players are evenly matched. Then there are 500games in which the winrates are 99% for and 500games in which the winrates are 1% for A. winrate of race A is therefore shown as (60*1000+99*500+1*500)/2000 = 55% instead of the true 60% balance. That's why stats with "lesser tournaments" will always be close to 50:50. As games in which one player is "outmatched" exist for both races and therefore the balance of those games is 50:50, the stats always tend to be closer to 50:50 as they should be from "Pro vs Pro" stats alone. Of course it remains to find out how big this influence really is. At least that's one approach to explain why aligulac (and other big statistics) never really has any outliers. the same can be said for foreign tournaments, where players like Hyun Jaedong Taeja Leenock dominate everyone else, and it seems more rare for a lot of the top korean protoss to attend. | ||
| ||
