Race statistics at NeoGame-i - Page 7
Forum Index > BW General |
STIMEY d okgm fish
Canada6140 Posts
| ||
Liquid`Ret
Netherlands4511 Posts
taling 7 pages about lT = imbalanced. LT = doesn't matter, positions is matter. 12z vs 9p = zerg imblaance (small margin) 12z vs 6p = zerg imbalance (small margin) 12z vs 3p = prootss imbalance ( by far ) 3z vs 12p = protoss imbalance ( by far) 3z vs 6p = zerg imbalance ( small margin) 3z vs 9p = zerg imbalance ( small margin ) 6z vs 12p / 3p = protoss imbalance ( small ) 6z vs 9p = balanced 9z vs 6 p = zerg imbalance 9z vs 12p / 3 p = protoss imbalance ( small) that's how LT is like, pretty much. and even thought small imbalance for p and z on some positions they are very very well capable of beating each other here. Vs a good protoss, every game is a fight, and it depends on what strategy u pick wether u win or lose 54.1% is 54 out of 100 games for zerg. On a server where random ppl play like neo-gamei that is a very good statistic imo.. Z and P is balanced. Either that or it's really damn close. | ||
Liquid`Nazgul
22426 Posts
On July 18 2004 13:28 TheBiShOp wrote: like frozen arbiter said , zerg have only won 3 mayor tournaments along bw history. If many protoss in korea have managed to win vs the best zergs in the world , u guys should know u can do it 2. And if u dont want this , just switch to z and lets make the neogamei ratio only z vs t t vs z and z vs z. don't be stupid LT is not used in any of the tournaments | ||
AnOther
565 Posts
| ||
Splinter
United States188 Posts
On July 18 2004 20:54 tfeign wrote: I did read the thread before posting. Sorry, you lose .Read the thread before you post next time. FA, Stats do not lie. Don't "theorycraft" AKA "you can beat zerg by playing better than the zerg" or "don't make mistakes." Well of course you can beat anyone by playing better than them. WTF is the point of saying that? The point is that PvZ is imbalanced. Statistics don't lie. 54.1% PvZ over 200000 games is significant if you know anything about statistics. Look at the stats. Don't theorycraft. Ask any toss player what their worst matchup is. If the majority doesn't say zerg I would never post in this thread again. You are correct though. Statistics don't lie. 54.1% of PvZs on NGI have been won by Zerg. That is the only fact these statistics support though. These statistics do not prove PvZ is imbalanced. You could use these statistics to hypothesize PvZ is imbalanced, but that far from proves anything. There are probably an infinite amount of variables that can go into determining the out come of a game that have NOTHING to do with how each race matches up against the other, so it is impossible to determine whether a matchup is imbalanced based on win/loss statistics. Also, a matchup being easier to play for one race compared to the other does not mean the matchup is imbalanced. Just something I want to bring up: Is a Matchup perfectly balanced if two players playing perfectly tie, or if two players playing perfectly have a 50% chance of winning? My guess is it is the later, as the former would make for a boring game. That being said, if a matchup is balanced so each race has a 50% chance of winning, that doesn't mean one of the races will actually win 50% of the games. It is like flipping a coin. Every time I flip a coin it has a 50% chance of landing on heads, and a 50% chance of landing on tails, but I could still flip a coin 20,000 times and end up with 15,000 tails. So, even if a matchup was balanced to give each race a 50% chance of winning, you would not be able to verify this by checking %s of actual games played. Anyway, at high level levels, SC games are decided by luck and guessing more than anything else. | ||
Beamo
France1279 Posts
On July 18 2004 00:21 tfeign wrote: Basically, the higher the number of experiments (games) are conducted, the closer and closer the % should merge to 50%. No It's -> the higher the number of expirements the more exact the result will be Ret: I don't agree with all your lt positioning blances for me 12-3 and 3-12 gives as much advantage to the toss as 9-6 or 6-9 does to the zerg. Having an over right over your base right at the start considerably limits your choices of build order even if zerg is at 6 and toss at 9 (yes I know 6 suxx ^^) | ||
Hoops
417 Posts
| ||
STIMEY d okgm fish
Canada6140 Posts
| ||
Liquid`Jinro
Sweden33719 Posts
On July 18 2004 20:54 tfeign wrote: Read the thread before you post next time. FA, Stats do not lie. Don't "theorycraft" AKA "you can beat zerg by playing better than the zerg" or "don't make mistakes." Well of course you can beat anyone by playing better than them. WTF is the point of saying that? The point is that PvZ is imbalanced. Statistics don't lie. 54.1% PvZ over 200000 games is significant if you know anything about statistics. Look at the stats. Don't theorycraft. Ask any toss player what their worst matchup is. If the majority doesn't say zerg I would never post in this thread again. PvZ is harder to get good at than PvT. Why? PvT macro and surround. PvZ TIMING+MICRO/MACRO. Timing = requires a LOT of games. Same with TvP as with PvZ except in a lesser degree. Look, imbalance only takes effect at the highest levels of BW - and there P's are doing pretty well. There's a bigger pool of zerg players to draw talent from (even though the neogamei % is pretty even, there are more zergs overall playing this game). I can't really explain myself due to my vocabulary not being extensive enough so this may look very confusing.. If you are an average player balance won't really affect you unless it's ridiculously bad (hm.. kind of like Gorky PvT). You'll only notice it very slightly, and it will be negated by differances in skill. It's a little easier for zerg, and LT at the very, very highest levels is not a fair map.. However, these stats are only the way they are becuase there's soooo many sucky tosses P vs Z. Urgh, I can't express myself.. Hopefully someone knows what I mean >< EDIT: What I'm saying is, unless they make really major changes (like tfeigns...) they'll not affect people at our level of play. | ||
radar14
United States1437 Posts
standard error = sq rt((0.54 * 0.46)/200000)) = 0.001114 = 0.1 percent 54% - 50% = 4% 4%/0.1% = 40 standard deviations If you know anything about statistics, 40 standard deviations pretty much undeniable proof that the null hypothesis is false. 3 standard deviations away would mean that, if the null was true, there would only be like a 5% chance that the given data would show up. And this is 40. So for whatever reason, Z > P from the neogamei sample. Whether or not that's attributable to P players just "not getting it" when playing Z or whatever, the disparity is there. | ||
NoName
United States1558 Posts
54% is not an independent standalone result. It is directly inversely related to 51.6% PvT. Players don't play random players... they play players at their rating, that will result in them winning 50% of all their games. All it means at a particular rating, against opponents of that rating, on average PvT will win 51.6% and PvZ will win 46%. Zs overall do seem to have slighly more wins, with about 51% overall win. T and P each win 49.5% of all games. This is probably due to Koreans (or Z players in general) more serious about the game and liking to play Z, more skilled, so as they are rising in ranking the Zs will get some extra wins. Saying 54% demonstrates inbalance is wrong. It just shows at a particularly rating level, P will lose more often Z, and P will beat T more often, but they will still win 50%. | ||
STIMEY d okgm fish
Canada6140 Posts
| ||
radar14
United States1437 Posts
All I am arguing is that PvZ is imbalanced, which has nothing to do with the sum of all matchups in the game. 40 standard deviations doesn't lie. You can't even look up the percentage for that data to occur if Z and P were equal, because past several standard deviations the percentage is essentially zero. | ||
Splinter
United States188 Posts
| ||
Beast_Bg
Bulgaria1623 Posts
| ||
aseq
Netherlands3929 Posts
This does indeed show that any random P player generally has more trouble facing a Z of his own level, than a T of his own level. However, these things cancel out quite well. (P's total might just be slightly under 50%). If the PvZ balance was (or could be) changed so that 'only' PvZ was affected, the 46% would rise, to say 49%. P would win more games, more than 50% total even. So they would move up in rank, until they were at about 50% again, but their race stats would not be 46,51.6,50 (pvp), but 49,52,50. Thus, their 3 percentages would be closer together, indicating there is a more even chance to win against any race. However, this would mean that if you make one race super strong, a player gets 90%,90% against other races, he would move up a great deal, until he was at 50%,50% again. So equal percentages do not mean everything. The only way to compare skills level for different races is at the top, the pro's. Here, we see a nice distribution of all races (top 20 pro's is very ok). So it would be best if we could get P from 51.6%,46% to 50%,50% without strengthening (is that a word) or weakening one of the races overall. The only 2 ways possible are to either make one unit from all races stronger or weaker, or to strengthen one unit (good vs Z) and weaken another (good vs T). The choice of unit would depend heavily on its use for both matchups. The probleam with the lurker and templar, the most commonly sought targets, is that these are both used a great deal vs. terrans as well, and so these are going to affect the other matchups too. Personally, I can't see a change in all 3 races (so not changing their overall strength compared to the other 2 races) that would have the desired effect. As TvZ seems very well-balanced, I would rather not change these two, it's very dangerous. So we should make P slightly stronger vs Z, but evenly slightly weaker vs T. Which unit is used vs T only, which unit is used vs Z only? My guess is that maybe the dragoon is more used vs T, and the archon or corsair vs Z. But those two don't seem to be the units that one would pick first. I wouldn't know what to change, and in that case, I'd rather change nothing at all than mess up the game. I don't believe P are less well played because there is a smaller player pool, there are still so many protoss that i don't think P is missing out on too many ideas. Sorry for the long post (and the english), this just shows that things are generally more complicated than just saying OMG FU IMBA LT NOOB Z SUX. | ||
STIMEY d okgm fish
Canada6140 Posts
| ||
NoNameLoser
United States1508 Posts
| ||
Liquid`Jinro
Sweden33719 Posts
| ||
STIMEY d okgm fish
Canada6140 Posts
| ||
| ||