|
On July 03 2012 02:09 Plansix wrote:Show nested quote +On July 03 2012 01:55 one-one-one wrote:Am I the only one in here that is annoyed by the poor level of knowledge about basic statistics? There is a lot of herpeteederp going on here about the interpretation of the presented data. Here is what this all means. It will be quite long, but this is why you should have been paying attention during the statistics lectures in high school. If you didn't do that you are definitely 100% not qualified to discuss in this thread. + Show Spoiler +see what I did there? Ok, so people have mentioned the "small" sample size being an issue. There seems to be some consensus that a larger sample is better. But how big does it have to be to give reliable answers? Imo. the creator of these graphs should do a more thorough presentation and also specify how reliable the numbers are. I will take the Korean TvZ numbers as an example. The data from June is based on 114 games. If you do the maths you will realize that there are 58 Terran wins and 56 Zerg wins. + Show Spoiler +58/114 = 0.50877.. 56/114 = 0.49122.. 56+58=114 Now, let us assume that there is an actual imbalance in the sense that over many games one of the races will win x% of them for players above some (high) skill level. Furthermore, we assume that the koreans were good enough to be above this skill level and that they played to win each game. Now we want to determine x using only a small sample of games. Imagine that we reversed time and played those 114 games again. It is unlikely that every game has the exact same outcome as before because of random stuff that might happen. + Show Spoiler +To make the problem more clear. Flip a coin twice. With a 50% likelihood you will get either tails-tails or heads-heads. From the sample size of 2 coin flips alone it is then hard to say that the coin-flip game is "balanced". Fortunately this problem is already solved in the world of statistics. We can calculate a so called confidence interval from the data. To do this, go to: http://www.mccallum-layton.co.uk/stats/ConfidenceIntervalCalcProportions.aspx and enter the numbers 114 as sample size and 50.9 as Observed Proportion. Set the confidence interval to 95% and calculate to get a confidence interval of 9.18%. The interpretation of this number is: with a certainty of 95% the number that we seek , x , is in the interval 50.9 ± 9.18 %. Now, this doesn't tell us very much does it ? It could be as bad as good (from a T perspective) as a 60.1% TvZ favor or as bad as a 41.7% disfavor. If we do the calculation with a larger sample size with data from TSL4 Code A qualifiers we get a sample size of 335 games with 139 Terran wins and 196 Zerg wins. The observed proportion is now: 139/335 = 41.5 %. Going back to the site above and entering these numbers gives us a 95% confidence interval of 5.28%. The number x would then lie in the interval 41.5 ± 5.28% , i.e. between 36.2% and 46.78% with a 95% certainty. + Show Spoiler +- But hey! You just proved that the game IS imbalanced. + Show Spoiler +- No N00B, I did not! l2maths !!! The problem is that my assumptions are a bit flawed. It is nice to idealize things to make the maths more clear, but in the real world Starcraft 2 is a constantly developing game. It only takes one game to revolutionize TvZ by abusing some timing or some other innovative play. People study each others games and learn from them. Only if these numbers persist over several months we can begin to talk about some kind of imbalance. And even if the numbers somehow finds their way into the 50±5 % interval that Blizzard wants it doesn't have to mean that the game is any better than today. It has been stressed so many times by clever individuals in this forum that one has to look subjectively on the games. Maybe Terran wins 90% of their games before minute 17 and Zerg shows equal figures thereafter which is a horrible kind of balance. After all, Blizzard wants an E-sport , right ? This means both a balanced and fair game and a nice viewer experience. I'll conceal some kind of balance whine in a spoiler to illustrate my point. Don't click this if you are a sensitive individual!! + Show Spoiler +Today, most TvZs I watch as a spectator are either some bio or marine tank bullshit that make the games go completely apeshit once brood lords are out on the field, or long stalemate games with mech where it is all about not getting your ravens fungaled. All ins are also quite common. Oh, don't forget to check out: http://en.wikipedia.org/wiki/Confidence_interval Wow, that is some amazing word smithing you did there. I like how you over complicate basic statistics in an effort to cook the numbers toward proving terran is underpowers. It falls directly in line with your previous post history in the "terran is having a hard time" threads in the past. I also like how you talk down to the readers and tell them they are not qualified to post about the stats unless they UNDERSTAND your findings. Highlight quotes below:"Furthermore, we assume that the koreans were good enough to be above this skill level and that they played to win each game. " - clearly we cannot assume that anyone in NA is playing to win. "It will be quite long, but this is why you should have been paying attention during the statistics lectures in high school. If you didn't do that you are definitely 100% not qualified to discuss in this thread." - Clearly we have been living our lives wrong all along. Time to leave my well playing job and go back to highschool. "Today, most TvZs I watch as a spectator are either some bio or marine tank bullshit that make the games go completely apeshit once brood lords are out on the field, or long stalemate games with mech where it is all about not getting your ravens fungaled. All ins are also quite common." - This has a ton to do with win rates."Maybe Terran wins 90% of their games before minute 17 and Zerg shows equal figures thereafter which is a horrible kind of balance. After all, Blizzard wants an E-sport , right ?" I don't really know what this means or what it has to do with over all win rates. If a terran loses a game before 17 minutes, does that no count either?This man has a future in cooking numbers. I work for some pretty large banks and they would love to employee someone like him.
I told you not to click that spoiler if you were sensitive , but you just had to didn't you?
It is beyond me how you can interpret the post as me trying to cook numbers? Didn't you read the second last spoiler?
I also explicitly said that some high school mathematics was needed. Said understanding would allow you to understand that my post has nothing to do with "cooking numbers". Clearly you didn't pay attention in high school.
For you employers sake I hope that your job doesn't require statistics.
What is clear is that you try to discredit me. It is a pretty lame attempt though which people who actually took the time to understand my post can see directly.
mods: I know I'm a bit out there on this one, but I was pissed off by him and wanted to defend myself. feel free to delete this post along with his attempt to discredit me if you want.
User was warned for this post
|
Are ~300 games a bit low with the tsl4 qualifier in kr, which are bo3 even in ro1?
Edit: I just read, that normally qualifiers are not taking into consideration. But with 300 games Kr is not really statistic relevant.
|
Lord_J
Kenya1085 Posts
On July 03 2012 02:22 Jaaaaasper wrote: These look supriseling balanced. I wonder if the zvt difference is due to the skill gap between Korean terrans and foreign terrans, sample size, or something else? All in all a lot better than i expected to see.
If you look at the race matchup statistics, they routinely vacillate between fairly extreme opposite positions from month-to-month, even when there were no intervening balance changes to account for the change. So, I don't think they are very persuasive evidence of imbalance or the lack thereof.
|
On July 03 2012 02:22 SniXSniPe wrote:Show nested quote +On July 03 2012 02:20 Plansix wrote:On July 03 2012 02:18 SniXSniPe wrote:On July 03 2012 02:15 Plansix wrote:On July 03 2012 02:13 SniXSniPe wrote: Why did they not include the TSL4 KR qualifiers? It would put Terran at around 41%ish win-ratio against Zerg. Because it is a qualifer, so anyone could sign up? I am pretty sure they did not include qualifers for other tournaments either. http://www.teamliquid.net/tournaments/admin/?action=bracket&id=3217Go through that list, and point out any subpar players for me. The fact is, every single TSL4 KR qualifier is completely stacked. If they don't include the qualifier for other tournaments, why would they include this one? Just because there are good players does not mean its a tournament. Because then you will have ignorant individuals claiming the statistics "prove" that match ups like TvZ are just fine, when in reality it is the opposite.
That is not how the process works. You can't just pick numbers that prove a specific point. They have to take the same results from the same sources every month. They cannot just add in qualifiers one month because favors a specific result.
|
just because the "balance in korea" looks better doesn't mean the matchup is particularly interesting to watch anymore.
|
On July 03 2012 01:32 Bluerain wrote:Show nested quote +On July 03 2012 00:59 Shiori wrote: I tire of seeing debate over these graphs, because it's always the same:
If something agrees with the balance whine of the day, everyone who plays the apparently overpowered race claims it's "metagame." If something disagrees, those same people turn right around and take the stats as gospel.
It's pretty simple: watch the games. TvZ and PvZ are both broken because Zerg in general is broken. Yes, Zerg was underpowered at release, but they haven't been for a long time. The other thing to consider is that in many respects, there were simply more top level Korean T/P players in tournaments than Zerg. Nestea is a notoriously weak traveller, and DRG is actually pretty new in the scheme of the Zerg scene; same with Symbol. There was a long period of time in which you had Nestea and then a big void of skill beneath him as far as Zerg went. Even now, I can only think of less than 10 truly top Zerg players, but I can think of at least 20 Terrans and probably 15 Protoss.
Tl;dr when MKP or Hero beat Moon or something it doesn't mean the matchup is balanced. dumbest/most biased post ever plus flame baiting. a biased zerg response would be that zerg is just UP and nestea is just way better so he can win while all other zerg players who are equal in skill to T/P players cannot win due to UP race. see how stupid heavily biased comments are? Except mine's based in fact. If you look at the most mechanically proficient/creative players, you get a lot of Terrans, a decent number of Protosses, and a few Zergs. It's not biased; it's just the way it is.
|
On July 03 2012 02:30 Shiori wrote:Show nested quote +On July 03 2012 01:32 Bluerain wrote:On July 03 2012 00:59 Shiori wrote: I tire of seeing debate over these graphs, because it's always the same:
If something agrees with the balance whine of the day, everyone who plays the apparently overpowered race claims it's "metagame." If something disagrees, those same people turn right around and take the stats as gospel.
It's pretty simple: watch the games. TvZ and PvZ are both broken because Zerg in general is broken. Yes, Zerg was underpowered at release, but they haven't been for a long time. The other thing to consider is that in many respects, there were simply more top level Korean T/P players in tournaments than Zerg. Nestea is a notoriously weak traveller, and DRG is actually pretty new in the scheme of the Zerg scene; same with Symbol. There was a long period of time in which you had Nestea and then a big void of skill beneath him as far as Zerg went. Even now, I can only think of less than 10 truly top Zerg players, but I can think of at least 20 Terrans and probably 15 Protoss.
Tl;dr when MKP or Hero beat Moon or something it doesn't mean the matchup is balanced. dumbest/most biased post ever plus flame baiting. a biased zerg response would be that zerg is just UP and nestea is just way better so he can win while all other zerg players who are equal in skill to T/P players cannot win due to UP race. see how stupid heavily biased comments are? Except mine's based in fact. If you look at the most mechanically proficient/creative players, you get a lot of Terrans, a decent number of Protosses, and a few Zergs. It's not biased; it's just the way it is. Sorry to burst your bubble, but you're actually a moron. Zing!
User was temp banned for this post.
|
On July 03 2012 00:54 platonichat wrote:Show nested quote +On July 03 2012 00:40 SeaSwift wrote:On July 03 2012 00:32 Shiori wrote:On July 03 2012 00:28 Roggay wrote:On July 03 2012 00:09 Shiori wrote: Incomplete set of data. Everyone who plays this game regularly knows that TvZ is utterly Zerg favoured, and that PvZ is utterly Zerg favoured if the Toss doesn't all-in. You don't need a graph to show this. Just look at the way you must play the game. Everyone who read TL regularly knows that you are a whiner too... Find me some professional Terrans (or honest Zergs) who think that TvZ isn't favouring Zerg right now. Besides, didn't you see the updated stats? Terran is at 41% if you factor those in. Considering Zerg was at 39% just last month, that actually isn't too bad at all considering how much whine was going round. edit: I was expecting something like a 25% winrate for Terran to justify all the whine. Nope, turns out that when Terran goes below 45% for the first time in the race's entire history it is enough to justify a forum explosion. Even forgetting about the fact that it was Terran at 39-41% last month, not Zerg, the fact that you expect a 25% winrate in TvZ in order to justify the MU being imbalanced shows you have no idea how this stuff works. This reddit post should be required reading for people to talk about these graphs. There is no way any race could get a 25% winrate especially not in the month after a change.
Yeah, ignore the mistake. My apologies.
I didn't say that. I didn't say that I "expect a 25% winrate in TvZ in order for the MU to be imbalanced". That is just putting words into my mouth. I said that there was so much whine - to the extent that the MU seemed to be literally impossible - and MKP etc were coming out with "10% winrate in practice" or 20% and the like in interviews. Comparatively, 41% is not bad at all.
|
wow looks like the win rates are pretty close in korea. im happy the game (win-rate wise) is as close to balanced as its been since release. happy times!
|
On July 03 2012 02:20 Plansix wrote:Show nested quote +On July 03 2012 02:18 SniXSniPe wrote:On July 03 2012 02:15 Plansix wrote:On July 03 2012 02:13 SniXSniPe wrote: Why did they not include the TSL4 KR qualifiers? It would put Terran at around 41%ish win-ratio against Zerg. Because it is a qualifer, so anyone could sign up? I am pretty sure they did not include qualifers for other tournaments either. http://www.teamliquid.net/tournaments/admin/?action=bracket&id=3217Go through that list, and point out any subpar players for me. The fact is, every single TSL4 KR qualifier is completely stacked. If they don't include the qualifier for other tournaments, why would they include this one? Just because there are good players does not mean its a tournament. They have more good players than ESV, if TLPD takes ESV stats from ro8 and up, why not do the same for Qualifiers? ESV has walkover even in ro8, which is way worse than code A qualifier.
Qualifier is also a big part of the balance. For example, if Terran race only have MKP, MMA, MVP and Zerg has 1000 players and Protoss has 500 players. As long as those 3 Terran keep holding 50% winrate, can we call it balance? Not counting qualifier works the same way. If there're only some Terran qualifying for tournaments against hordes of Zerg and Protoss, I don't think it is balanced. The fact that those Terran managing to qualify shows that they are good enough to hold 50% easily against the other races(minus the fact that they win mirror MU). So in paper and TLPD, Terran doesn't look bad, but it's deeper than that.
|
On July 03 2012 02:24 one-one-one wrote:Show nested quote +On July 03 2012 02:09 Plansix wrote:On July 03 2012 01:55 one-one-one wrote:Am I the only one in here that is annoyed by the poor level of knowledge about basic statistics? There is a lot of herpeteederp going on here about the interpretation of the presented data. Here is what this all means. It will be quite long, but this is why you should have been paying attention during the statistics lectures in high school. If you didn't do that you are definitely 100% not qualified to discuss in this thread. + Show Spoiler +see what I did there? Ok, so people have mentioned the "small" sample size being an issue. There seems to be some consensus that a larger sample is better. But how big does it have to be to give reliable answers? Imo. the creator of these graphs should do a more thorough presentation and also specify how reliable the numbers are. I will take the Korean TvZ numbers as an example. The data from June is based on 114 games. If you do the maths you will realize that there are 58 Terran wins and 56 Zerg wins. + Show Spoiler +58/114 = 0.50877.. 56/114 = 0.49122.. 56+58=114 Now, let us assume that there is an actual imbalance in the sense that over many games one of the races will win x% of them for players above some (high) skill level. Furthermore, we assume that the koreans were good enough to be above this skill level and that they played to win each game. Now we want to determine x using only a small sample of games. Imagine that we reversed time and played those 114 games again. It is unlikely that every game has the exact same outcome as before because of random stuff that might happen. + Show Spoiler +To make the problem more clear. Flip a coin twice. With a 50% likelihood you will get either tails-tails or heads-heads. From the sample size of 2 coin flips alone it is then hard to say that the coin-flip game is "balanced". Fortunately this problem is already solved in the world of statistics. We can calculate a so called confidence interval from the data. To do this, go to: http://www.mccallum-layton.co.uk/stats/ConfidenceIntervalCalcProportions.aspx and enter the numbers 114 as sample size and 50.9 as Observed Proportion. Set the confidence interval to 95% and calculate to get a confidence interval of 9.18%. The interpretation of this number is: with a certainty of 95% the number that we seek , x , is in the interval 50.9 ± 9.18 %. Now, this doesn't tell us very much does it ? It could be as bad as good (from a T perspective) as a 60.1% TvZ favor or as bad as a 41.7% disfavor. If we do the calculation with a larger sample size with data from TSL4 Code A qualifiers we get a sample size of 335 games with 139 Terran wins and 196 Zerg wins. The observed proportion is now: 139/335 = 41.5 %. Going back to the site above and entering these numbers gives us a 95% confidence interval of 5.28%. The number x would then lie in the interval 41.5 ± 5.28% , i.e. between 36.2% and 46.78% with a 95% certainty. + Show Spoiler +- But hey! You just proved that the game IS imbalanced. + Show Spoiler +- No N00B, I did not! l2maths !!! The problem is that my assumptions are a bit flawed. It is nice to idealize things to make the maths more clear, but in the real world Starcraft 2 is a constantly developing game. It only takes one game to revolutionize TvZ by abusing some timing or some other innovative play. People study each others games and learn from them. Only if these numbers persist over several months we can begin to talk about some kind of imbalance. And even if the numbers somehow finds their way into the 50±5 % interval that Blizzard wants it doesn't have to mean that the game is any better than today. It has been stressed so many times by clever individuals in this forum that one has to look subjectively on the games. Maybe Terran wins 90% of their games before minute 17 and Zerg shows equal figures thereafter which is a horrible kind of balance. After all, Blizzard wants an E-sport , right ? This means both a balanced and fair game and a nice viewer experience. I'll conceal some kind of balance whine in a spoiler to illustrate my point. Don't click this if you are a sensitive individual!! + Show Spoiler +Today, most TvZs I watch as a spectator are either some bio or marine tank bullshit that make the games go completely apeshit once brood lords are out on the field, or long stalemate games with mech where it is all about not getting your ravens fungaled. All ins are also quite common. Oh, don't forget to check out: http://en.wikipedia.org/wiki/Confidence_interval Wow, that is some amazing word smithing you did there. I like how you over complicate basic statistics in an effort to cook the numbers toward proving terran is underpowers. It falls directly in line with your previous post history in the "terran is having a hard time" threads in the past. I also like how you talk down to the readers and tell them they are not qualified to post about the stats unless they UNDERSTAND your findings. Highlight quotes below:"Furthermore, we assume that the koreans were good enough to be above this skill level and that they played to win each game. " - clearly we cannot assume that anyone in NA is playing to win. "It will be quite long, but this is why you should have been paying attention during the statistics lectures in high school. If you didn't do that you are definitely 100% not qualified to discuss in this thread." - Clearly we have been living our lives wrong all along. Time to leave my well playing job and go back to highschool. "Today, most TvZs I watch as a spectator are either some bio or marine tank bullshit that make the games go completely apeshit once brood lords are out on the field, or long stalemate games with mech where it is all about not getting your ravens fungaled. All ins are also quite common." - This has a ton to do with win rates."Maybe Terran wins 90% of their games before minute 17 and Zerg shows equal figures thereafter which is a horrible kind of balance. After all, Blizzard wants an E-sport , right ?" I don't really know what this means or what it has to do with over all win rates. If a terran loses a game before 17 minutes, does that no count either?This man has a future in cooking numbers. I work for some pretty large banks and they would love to employee someone like him. I told you not to click that spoiler if you were sensitive , but you just had to didn't you? It is beyond me how you can interpret the post as me trying to cook numbers? Didn't you read the second last spoiler? I also explicitly said that some high school mathematics was needed. Said understanding would allow you to understand that my post has nothing to do with "cooking numbers". Clearly you didn't pay attention in high school. For you employers sake I hope that your job doesn't require statistics. What is clear is that you try to discredit me. It is a pretty lame attempt though which people who actually took the time to understand my post can see directly. mods: I know I'm a bit out there on this one, but I was pissed off by him and wanted to defend myself. feel free to delete this post along with his attempt to discredit me if you want.
+ Show Spoiler +but his math is on as best as I can tell. At least barring a source error. And all he really did was show that math by itself leaves the balance question ambiguous, especially with small sample sizes over short periods of time. Care to explain what cooking you saw?
|
On July 03 2012 02:32 The_Stampede wrote:Show nested quote +On July 03 2012 02:30 Shiori wrote:On July 03 2012 01:32 Bluerain wrote:On July 03 2012 00:59 Shiori wrote: I tire of seeing debate over these graphs, because it's always the same:
If something agrees with the balance whine of the day, everyone who plays the apparently overpowered race claims it's "metagame." If something disagrees, those same people turn right around and take the stats as gospel.
It's pretty simple: watch the games. TvZ and PvZ are both broken because Zerg in general is broken. Yes, Zerg was underpowered at release, but they haven't been for a long time. The other thing to consider is that in many respects, there were simply more top level Korean T/P players in tournaments than Zerg. Nestea is a notoriously weak traveller, and DRG is actually pretty new in the scheme of the Zerg scene; same with Symbol. There was a long period of time in which you had Nestea and then a big void of skill beneath him as far as Zerg went. Even now, I can only think of less than 10 truly top Zerg players, but I can think of at least 20 Terrans and probably 15 Protoss.
Tl;dr when MKP or Hero beat Moon or something it doesn't mean the matchup is balanced. dumbest/most biased post ever plus flame baiting. a biased zerg response would be that zerg is just UP and nestea is just way better so he can win while all other zerg players who are equal in skill to T/P players cannot win due to UP race. see how stupid heavily biased comments are? Except mine's based in fact. If you look at the most mechanically proficient/creative players, you get a lot of Terrans, a decent number of Protosses, and a few Zergs. It's not biased; it's just the way it is. Sorry to burst your bubble, but you're actually a moron. Zing! Feel free to back up your argument with facts rather than sniveling. I don't even play Terran and I can see that they have the most mechanically skilled players.
|
Looking forward to seeing all Terran players turning this thread into another wine and bitch thread.
|
the ZvT graph seems to have corrected itself beautifully. First time in two years where Z has a higher win percentage than T.
|
On July 03 2012 02:36 Shiori wrote:Show nested quote +On July 03 2012 02:32 The_Stampede wrote:On July 03 2012 02:30 Shiori wrote:On July 03 2012 01:32 Bluerain wrote:On July 03 2012 00:59 Shiori wrote: I tire of seeing debate over these graphs, because it's always the same:
If something agrees with the balance whine of the day, everyone who plays the apparently overpowered race claims it's "metagame." If something disagrees, those same people turn right around and take the stats as gospel.
It's pretty simple: watch the games. TvZ and PvZ are both broken because Zerg in general is broken. Yes, Zerg was underpowered at release, but they haven't been for a long time. The other thing to consider is that in many respects, there were simply more top level Korean T/P players in tournaments than Zerg. Nestea is a notoriously weak traveller, and DRG is actually pretty new in the scheme of the Zerg scene; same with Symbol. There was a long period of time in which you had Nestea and then a big void of skill beneath him as far as Zerg went. Even now, I can only think of less than 10 truly top Zerg players, but I can think of at least 20 Terrans and probably 15 Protoss.
Tl;dr when MKP or Hero beat Moon or something it doesn't mean the matchup is balanced. dumbest/most biased post ever plus flame baiting. a biased zerg response would be that zerg is just UP and nestea is just way better so he can win while all other zerg players who are equal in skill to T/P players cannot win due to UP race. see how stupid heavily biased comments are? Except mine's based in fact. If you look at the most mechanically proficient/creative players, you get a lot of Terrans, a decent number of Protosses, and a few Zergs. It's not biased; it's just the way it is. Sorry to burst your bubble, but you're actually a moron. Zing! Feel free to back up your argument with facts rather than sniveling. I don't even play Terran and I can see that they have the most mechanically skilled players. What evidence could you POSSIBLY use to make this claim?
|
On July 03 2012 01:51 mrtomjones wrote:Show nested quote +On July 03 2012 01:43 EnE wrote:On July 02 2012 23:04 Kenshi235 wrote: I don't believe the TvZ winrate in Korea ~50/50. I'm calling bs on this. I'd like to know who regularly posts this and what games these are derived from.
I finally found out sample size is only 114 for KR TvZ and excludes Code A qualifers, TSL4 qualifiers, VS 996 sample size of international graph. I'm sorry, but I don't believe Koreans adapted to patch perfectly.
Code A + TSL qualifers are 220 sets alone. Good job lying with your statistics. I added TSL and Code A qualifers in from data further in post. Sample size is 334 and TvZ winrate is 41.5% for T. This is why you don't believe everything you hear. THIS!Please lock this thread now lol Evidence must be wrong! It doesn't agree with what I know for a fact! If you listened to certain shows in the last while you would realize that Korean Terrans HAVE adapted quite a bit already. If you want to look at balance you should never look at international too much. Korea also ALWAYS adapts faster than International scenes do. Code A qualifiers shouldn't be included due to the non Pro players in there and the people who just aren't very good playing in it. TSL should probably be included but yah.
"Don't include korean top masters and gm in win rates because you can't use those to represent balance at a high level"
You'll never be a code B player.
|
oh terrans with highest winrates in korea. surprise!
|
On July 03 2012 02:41 DeCoder wrote: the ZvT graph seems to have corrected itself beautifully. First time in two years where Z has a higher win percentage than T.
No, if you read my post on the previous page you will see that the only thing you can say from whose numbers is something along the lines of: With a 95% certainty we can say that the win-rate over many games is between 42 and 60% for Terran.
edit: your interpretation is the one that people tend to make naturally, but if you examine it further the data is inconclusive.
|
On July 03 2012 02:45 one-one-one wrote:Show nested quote +On July 03 2012 02:41 DeCoder wrote: the ZvT graph seems to have corrected itself beautifully. First time in two years where Z has a higher win percentage than T. No, if you read my post on the previous page you will see that the only thing you can say from whose numbers is something along the lines of: With a 95% certainty we can say that the win-rate over many games is between 42 and 60% for Terran. edit: your interpretation is the one that people tend to make naturally, but if you examine it further the data is inconclusive.
Clearly, we can tell that some number was reached and some % of games were won by terrans. It is likely between 25% and 75%. Or maybe it is just 50%.
|
On July 03 2012 02:51 Plansix wrote:Show nested quote +On July 03 2012 02:45 one-one-one wrote:On July 03 2012 02:41 DeCoder wrote: the ZvT graph seems to have corrected itself beautifully. First time in two years where Z has a higher win percentage than T. No, if you read my post on the previous page you will see that the only thing you can say from whose numbers is something along the lines of: With a 95% certainty we can say that the win-rate over many games is between 42 and 60% for Terran. edit: your interpretation is the one that people tend to make naturally, but if you examine it further the data is inconclusive. Clearly, we can tell that some number was reached and some % of games were won by terrans. It is likely between 25% and 75%. Or maybe it is just 50%.
maybe. that is not the point. the point is that the data is inconclusive.
now, would you enlighten us and tell me in what way I was cooking numbers?
3 people besides me has already called you out on it.
edit: and it is likely NOT between 25% and 75%. you just cooked those numbers up, lol.
|
|
|
|