EPTAs far back as last November, when terran was quite favored in TvZ
After ghost changes, MMA is going to lose more.
Top Tier Korean ZvT and TvZ TLPD statistics - Page 5
| Forum Index > SC2 General |
|
SafeAsCheese
United States4924 Posts
As far back as last November, when terran was quite favored in TvZ After ghost changes, MMA is going to lose more. | ||
|
McFeser
United States2458 Posts
| ||
|
MercilessMonkey
Canada150 Posts
On March 16 2012 12:51 HyperionDreamer wrote: Yep. The study cited in the OP pertains to a specific type of stats testing, called cluster analysis. Maybe read up on it a bit before you cite it as valid, OP. You're talking about simple testing for type 1/2 statistical errors, so you would need a much larger sample size. I did a post a while ago doing rigid scientific statistical analysis on korean matchup percentages, and I think even a sample size of ~200 games rendered a ~7% difference statistically irrelevant. http://en.wikipedia.org/wiki/Cluster_analysis Edit: It was a sample size of 130, and an ~8% statistical difference. This was rendered statistically insignificant using standard p-level analysis. Here's the link to my analysis. http://www.teamliquid.net/forum/viewmessage.php?topic_id=317114¤tpage=12#226 Glad people beat me to it. But OP is "publishing works" so he can't be wrong, right? =/ Anyhow, as for what he's doing, the above two are basically right. Your confidence interval is going to be gigantic, so unless you're testing at something silly like a 0.5 confidence level all of your results are going to be statistically insignificant with your sample size. Probably shouldn't call other people out on not understand statistics. | ||
|
neoghaleon55
United States7435 Posts
On March 16 2012 13:01 McFeser wrote: Happy's TvZ is 83% by the way. 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. | ||
|
Ireniicus
United Kingdom374 Posts
On March 16 2012 11:41 hunts wrote: 64% may be good for a sub par player, but nnot for a champion like DRG or nestea This | ||
|
Charger
United States2405 Posts
| ||
|
dgwow
Canada1024 Posts
| ||
|
Veldril
Thailand1817 Posts
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. On March 16 2012 11:41 hunts wrote: 64% may be good for a sub par player, but nnot for a champion like 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). | ||
|
Mr. Nefarious
United States515 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. Wow, thank you. | ||
|
Hipsv
135 Posts
On March 16 2012 12:52 neoghaleon55 wrote: Where are you getting this "tilting the statistics" and "exaggerating"? All this information is from TLPD...if you have a problem with them, tell teamliquid. and I'm not exaggerating , 5% is very lenient, if you look at the numbers it's actually closer to 6-7%...but people like nice numbers like 5  You are tilting the statistics by including patches long ago where Terran was at an advantage over Zerg, this is a given fact. You attempt to disguise your intent through ignorance, but in order for someone to spend as much time as you did to make a thread about this you have a vested interest in the public response. The statistics mean nothing because your assumptions (whether you intended to assume them or not) don't match the data that is included. | ||
|
blade55555
United States17423 Posts
Not this. If you look at BW for instance Jaedong has a 68% and Flash 71% both are amazing players and their not that much further then 64%. Most don't have above 64% and if they do it's not by much. | ||
|
NoctemSC
United States771 Posts
I agree with you on the discrepancy difference between TvZ and ZvT and I'll explain my reason. First off, never compare MMA to anyone else. His TvZ is ridiculous and he is not a good example. Terran know they are at a disadvantage vs Zerg late game. Because of this Terran are forced to all in. Zerg are weak vs all in play at all levels because they want to go macro heavy, get that fast expand etc. What Terran are doing is simply exploiting the greedy zerg mentality. It would be easy for Zerg to adapt to a slightly less greedy build order and defend as I've seen it done quite well. It doesn't come down to Terran being stronger, it comes down to Terran exploiting a self made weakness that many greedy zergs have set up. You can't expect to have the advantage late AND early game. That would be unrealistic. As a Turtle Terran I'd gladly trade my aggressive early game for a powerful late game. The fact is that Zerg have awesome early game aggression as well, provided they know what they're doing. Simply look at Leeknock. | ||
|
zefreak
United States2731 Posts
| ||
|
neoghaleon55
United States7435 Posts
On March 16 2012 13:06 Hipsv wrote: You are tilting the statistics by including patches long ago where Terran was at an advantage over Zerg, this is a given fact. You attempt to disguise your intent through ignorance, but in order for someone to spend as much time as you did to make a thread about this you have a vested interest in the public response. The statistics mean nothing because your assumptions (whether you intended to assume them or not) don't match the data that is included. I'm just giving you numbers from teamliquid's very own database. You may suggest changes to how TLPD compile their stats in the website feedback forums. | ||
|
Let it Raine
Canada1245 Posts
is this correct? the first time i saw DRG (was in the message from korea TL topic, but that's not tournaments) was the GSTL where he was dominating everyone | ||
|
Oboeman
Canada3980 Posts
On March 16 2012 13:06 Hipsv wrote: You are tilting the statistics by including patches long ago where Terran was at an advantage over Zerg, this is a given fact. You attempt to disguise your intent through ignorance, but in order for someone to spend as much time as you did to make a thread about this you have a vested interest in the public response. The statistics mean nothing because your assumptions (whether you intended to assume them or not) don't match the data that is included. I don't think either DRG or MMA were competing in the GSL yet on Kulas and Steppes and Jungle Basin, or with rax before depot, or 3 range roaches, or low health hatcheries, or fast medivacs, or anything like that. I think by the time they showed their faces things were pretty stable. The "dark ages" shouldn't really effect these. | ||
|
ajabberwok
United States59 Posts
http://en.wikipedia.org/wiki/Margin_of_error#Calculations_assuming_random_sampling error on this type of measurement = sqrt( p*(1-p) / n ) where p = percentage of wins, and n is the total sample size (total number of games). So, for example: MMA vs Z: p=0.8077, n=26, so error = 0.077 DRG vs T: p=0.6400, n=50, so error = 0.068 Their win percentage is further apart than the error on each measurement - so there is a statistically significant difference. | ||
|
Sideburn
United States442 Posts
On March 16 2012 11:14 neoghaleon55 wrote: Edit 1: The optimal minimum sample size is 20. Above 20, the n value does not relevantly contribute over all (n-1) to the statistics. I'm sure you remember from your AP stats class and college. The statistics presented in the OP are greater than 20 sample sizes and thus are relevant. Edit 2: Maybe presenting that article wasn't such a good idea as it only confuses people more. Let me try to explain this in easier terms to understand. So how about this...http://en.wikipedia.org/wiki/Standard_deviation lol Statistics is very dependent on standard deviations which accounts for your confidence interval. Standard deviation (SI) uses an (N-1) factor, which contributes less and less as N gets larger. at 20 or above, N-1 is seen as negligible in mathematics terms. I actually use 20 or greater in my research and published works as well... it's quite well known. Thank god for you. Honestly, I nearly choked up reading the end of your post. I cannot tell you how incredibly frustrated I get when people simply dismiss things claiming "The sample size is too small" without having ANY idea of what that means. In fact, I have often seen people on this forum claim "The sample size is too small" when it isn't a sample at all. It's actually population data. Now... you can't just say that 20 is fine the way you are. In fact it really isn't as someone pointed out earlier in this thread. Still a devotion to actual statistics is a really really important thing. I think that innumeracy is a huge problem in this community and really society in general. | ||
|
neoghaleon55
United States7435 Posts
On March 16 2012 13:11 zefreak wrote: Why is anyone taking this thread seriously? Have you guys not seen his other threads? Why anyone would want to learn statistics from the creator of the SC2 fanfic threads is beyond me. If he has actually published papers in peer reviewed journals with the understanding he has shown in this thread I would be shocked. I like how you go for personal attacks rather than addressing the data. I'm published in the journal of physical chemistry You may look up A. Montenegro, J. S. A. Ishibashi, P. Lam, Z. Li: “Kinetics Study of Reaction of Pinenes with Hydroxyl Radical at 1–8 Torr and 240-340 K Using the Relative Rate/Discharge Flow/Mass Spectrometry Method". | ||
|
Vindicare605
United States16121 Posts
| ||
| ||
