|
This discussion will lead nowhere untill the community can agree upon common definitions of an over-/under-powered race, the magnitude one race has to outperform other races to be considered overpowered, the persitance of overperformance of a race to be considered overpowered, the way in which a race outperform other races to be considered overpowered (winratio, grandmasterpresence to race distribution-quote ect.), where such outperformance should be measured and considered (LANS / tournaments / ladder ect.), which players constitutes a representative sample to measure imbalance (high level / top level / all levels?) ect.
Before you so this, the discussion will not amount to anything and nothing productive will ever come from it. "random" statistics will continue to be presented and people will talk over eachother for a few pages without coming to any conclusion.
A great deal of data is available from the top tiers of player so there's no excuse for anyone serious about trying to find imbalance to start on this project. The key point is to be honest about what you find and what you don't find and not pick sample data that happens to suit your point of view. That's as dishonest as it is common 
|
United States7483 Posts
I would argue it is a result of Terran being the most cheese proof of the three races, and having the largest variety of opening strategies, as well as being the hardest to scout and having the easiest scouting. It's not necessarily that they're stronger, but just have more viability in a system as volatile as ladder.
The ladder is volatile because every game is just that: one game. You might know your opponent in Grand Masters, but it doesn't matter, which means your foe could be doing anything from ridiculous builds for learning to really standard play etc.
|
With minimal statistics comprehension, you should realize that a because there is a majority of P players in GM, it means they play more against each other, therefore averaging their own race's win rate.
You should have considered P's ratio against T and Z in your analysis.
|
What your research shows is that right now the best Terran's are able to reach and stay at the top of their ladder better then the top Protoss and Zerg. Essentially they are more Stable.
I think this is because the TvT match-up is the most stable and generally allows the better player to win, where as right now PvP is a guessing game, and ZvZ is always crazy.
If you look at BW players, the ones that stay at the top of the rankings always dominate their mirror match-up. The best BW ZvZ, TvT and PvP in the last few years have usually been Jaedong, Flash, and Bisu respectively. The correlation between staying at a high ranking and having a good mirror win rate is an often overlooked factor.
Right now it's hard for P and Z to have a high mirror win rate because the match-ups are not as stable as TvT.
|
I've got some stats for you:
My research indicates that the least popular race in the Bronze League is Zerg. Doesn't that mean that Zerg is terribly OP if it's less likely to be horrible?
|
100 is obviously too small to draw any conclusion. from what i saw, top 1,000 players, protoss has the most number.
|
On April 26 2011 05:01 ehalf wrote:
100 is obviously too small to draw any conclusion. from what i saw, top 1,000 players, protoss has the most number.
Except the skill divide between the top ~300 and the rest is bigger than the divide between Master's and Diamond
|
This is not a good statistical analysis I think your just trying to look for a pattern. All your are doing is taking numbers and showing correlations that prove absolutely nothing. Its like almost saying that more crimes occur in the summer and more ice cream sales also occur in the summer. So due to the statistics crime and ice cream sales have a direct correlation.
It seems like your just looking into trolling and causing a muck in the community for your own selfish personal reasons. Everyone knows statistics doesn't mean anything on ladder and the tournaments show that too. If statistics hold true to their meaning then the top players on ladder should have been winning tournaments left and right which is not the case.
The game is still being developed and statistics don't mean crap on something that is being developed. For all we know we could be playing the game completely wrong 1-2 years down the road from now.
Scientific evidence does not work on something as volatile as sc2 ladder nor tournaments so please people need to just can it with the statistics.
|
Terran is probably the easiest race to play, ofc there are minor balance issues but that is most likely the biggest contributor to terran dominance.
|
On April 26 2011 03:40 epoc wrote:Show nested quote +On April 26 2011 03:29 4of8 wrote:On April 26 2011 02:40 KillerPlague wrote:
There currently is 1638 players listed in GM. The distribution is:
2.6% random
38% protoss
30.3% terran
29% zerg
being a protoss player myself, it's a little hard to ignore how many protoss actually made it in to GM. considering this is the top 200 cumulative for each region it feels like your statistics are biased to make terran seem worse than they are. however, this may be due to the simple fact that more people play protoss? as far as terran dominating the ladder, i think it is due to the relative safety they have. zerg players must scout like hell, protoss players must adapt their composition, terran players can do the same build over and over again vs either race with relative safety. things like salvageable bunkers add to this, as many good players will opt for them when they are not sure. The point is under this 1638 players are players from all Servers and if you look at each server you will see Protoss is dominating the weaker Servers like China ( game was released only some weeks ago => obvious one of the weaker servers at the moment) or Latin America.. but in Europe and Korea terran takes the lead. But this imbalance discussions are really funny. Like in the recent past everybody is saying toss is dominating the big tourneys and then I take a look on the CODE S and CODE A results or the GSL WC and it is not really looking good for toss. And then comes always the quoting of the progamers, but if you haven't mentioned it yet, they are always saying their race is up. Look at guys like Idra, he would even say Zerg is up if they would have 10 Zergplayers in the Top Ten. For some of these guys it might be only the fact, that they are searching for a easy excuse. But I think the main reason is, Blizzard said on the Blizzcon, the opinion of the progamers would be one part of balancing the game. So ask yourself would you really tell the public, if you are thinking your race is op? Even if we would have a p-value under 5% for this stats. This wouldn't mean Toss is up or Terran op. This game is still very young and the patch still new... the only thing we could say that certain strats are at the moment strong or not strong. And with this stats it seems like terrans have at the moment maybe the best working strats for these ladder maps. But nobody knows if this will change in the next weeks even without a patch. Yes Idra would say that. When has Idra been wrong about balance?
Pretty much every time he has spoken about it except when concerning pre-nerf Siege Tanks (which I dont think anyone would argue were balanced, except maybe Avilo) and early map balance (derp). But this goes for many many many other people. MVP thinks Terran is weak, and Nestea thinks Zerg is weak. What a twist! I think Im spotting a tendency here that indicates what a player playing a certain race will say about said race's balance.
Personally I stick to the Law of Artosis: When dealing with a percieved balance issue, refer to the three golden rules:
Terran is OP
Zerg is UP
Protoss is gay
and your question will be answered... somewhat. This thread is therefore redundant. Mostly because all it really highlights is that the ladder is showing incredibly balanced statistics. Of course the ladder is not an indication of anything, unless it indicates a point you want to bring across.
|
Only looking at the top 100 is absolutely silly. The pros themselves say that the ladder rankings matter less and less the farther up you go. Many of the best players are not in the top 100:
MC
Nada
San
Anypro
NesTea
Select
Mondragon
Clearly many of these guys would be in the actual best 100 players in the world. Point is that the top 100 on ladder is not at all accurate of the actual best 100 players and is pretty much insignificant.
|
On April 26 2011 02:02 Shaetan wrote:
You can't claim to do statistical analysis then just throw out numbers and claim that X is true, you have to do the analysis.
This...
On April 26 2011 02:25 dogmeatstew wrote:
All the win %'s are very close. The average number of points is very close, 15 points isn't a lot and as you've already stated, a large amount of top 10 is terrans which would easily skew this.
In short I'm not sure how these numbers give you the impression that terran is dominating in every area, a 2% higher win ratio isn't statistically significant even across 1638 players and all in all this last set of data actually looks pretty even across the board to me.
...and this.
Let me start by saying that I don't think it's very useful to try and statistically analyze the current state of balance based off of the Blizzard ladder. I just want to make a point that if you're going to attempt using statistics, you should actually use statistical analysis and not just look at numbers and try and determine if patterns exist -- this doesn't work and this is why the field of statistics exists.
I'm not a statistician, just a lowly science grad student, but I did some really, really simple stat analysis on the North American Grandmaster league statistics (it would take a really long time to gather the data from every league into Excel). If you just look at the data (like the OP) did, it seems like there are some differences here:
Win% by race:
P: 59.0% (15,632)
T: 61.4% (10,083)
Z: 59.7% (11,163)
R: 55.4% (1,333)
But you can't really tell if there is or not without actually analyzing the data. So I ran a one-way ANOVA on the NA GM league, comparing win% between the races. All the data were pulled from SC2Ranks. The data weren't normal, so a square root transformation was applied [automatically, by the program] to meet the assumptions of normality for ANOVA. The assumption of homogeneity of variances was also met (p=0.430).
Here is the descriptive statistics table generated:
![[image loading]](http://i.imgur.com/LkdRn.png)
From just this table, you can see that the 95% CIs overlap between every race. Note that the mean values are the actual values, not the sqrt transformed values (not sure why the program does this).
The ANOVA table:
![[image loading]](http://i.imgur.com/DvH3v.png)
This shows that there is a statistically significant (p=0.043) difference overall between the group means, but it doesn't tell us where the differences are.
This is where post-hoc tests are important, as they lets us look at comparisons between multiple groups:
Link for Big! Here, we have two post-hoc test results. In this table 1=P, 2=T, 3=Z, 4=R.
We can see that in the Tukey test, there is a marginally significant difference between T and P win%s, and the Least Significant Difference test shows that there is a significant difference between T and P. Essentially, what we glean from this entire test is that the only real differences in win% are between Terran and Protoss, and that difference is pretty small.
To visualize the actual comparisons, here is a bar graph with ERROR BARS!
![[image loading]](http://i.imgur.com/LfocY.png)
TL;DR: Analysis of variance shows that the only difference between NA GM win%s is in T and P -- and it's not very big. Don't make threads about taking 'mathematical' approaches or statistical comparisons if you're not going to actually use statistics. Looking at numbers =/= statistics.
|
On April 26 2011 05:49 Pelican wrote:Show nested quote +On April 26 2011 02:02 Shaetan wrote:
You can't claim to do statistical analysis then just throw out numbers and claim that X is true, you have to do the analysis. This... Show nested quote +On April 26 2011 02:25 dogmeatstew wrote:
All the win %'s are very close. The average number of points is very close, 15 points isn't a lot and as you've already stated, a large amount of top 10 is terrans which would easily skew this.
In short I'm not sure how these numbers give you the impression that terran is dominating in every area, a 2% higher win ratio isn't statistically significant even across 1638 players and all in all this last set of data actually looks pretty even across the board to me. ...and this. Let me start by saying that I don't think it's very useful to try and statistically analyze the current state of balance based off of the Blizzard ladder. I just want to make a point that if you're going to attempt using statistics, you should actually use statistical analysis and not just look at numbers and try and determine if patterns exist -- this doesn't work and this is why the field of statistics exists. [...] Show nested quote +I'm not a statistician, just a lowly science grad student, but I did some really, really simple stat analysis the North American Grandmaster league statistics (it would take a really long time to gather the data from every league into Excel). If you just look at the data (like the OP) did, it seems like there are some differences here: Win% by race: P: 59.0% (15,632) T: 61.4% (10,083) Z: 59.7% (11,163) R: 55.4% (1,333) But you can't really tell if there is or not without actually analyzing the data. So I ran a one-way ANOVA on the NA GM league, comparing win% between the races. All the data were pulled from SC2Ranks. The data weren't normal, so a square root transformation was applied [automatically, by the program] to meet the assumptions of normality for ANOVA. The assumption of homogeneity of variances was also met (p=0.430). Here is the descriptive statistics table generated: From just this table, you can see that the 95% CIs overlap between every race. Note that the mean values are the actual values, not the sqrt transformed values (not sure why the program does this). The ANOVA table: This shows that there is a statistically significant (p=0.043) difference overall between the group means, but it doesn't tell us where the differences are. This is where post-hoc tests are important, as they lets us look at comparisons between multiple groups: Link for Big! Here, we have two post-hoc test results. In this table 1=T, 2=P, 3=Z, 4=R. We can see that in the Tukey test, there is a marginally significant difference between T and P win%s, and the Least Significant Difference test shows that there is a significant difference between T and P. Essentially, what we glean from this entire test is that the only real differences in win% are between Terran and Protoss, and that difference is pretty small. To visualize the actual comparisons, here is a bar graph with ERROR BARS! [...] TL;DR: Analysis of variance shows that the only difference between NA GM win%s is in T and P -- and it's not very big. Don't make threads about taking 'mathematical' approaches or statistical comparisons if you're not going to actually use statistics. Looking at numbers =/= statistics.
Thank god someone else shares my opinion. There has been a proliferation of pseudo-statistics threads recently by people who clearly have little statistical or mathematical background. All these threads do is troll people by generating meaningless debates. Statistics and applied math undergrad here.
|
On April 26 2011 05:46 SqueamishCow wrote:
Only looking at the top 100 is absolutely silly. The pros themselves say that the ladder rankings matter less and less the farther up you go. Many of the best players are not in the top 100:
MC
Nada
San
Anypro
NesTea
Select
Mondragon
Clearly many of these guys would be in the actual best 100 players in the world. Point is that the top 100 on ladder is not at all accurate of the actual best 100 players and is pretty much insignificant.
This has been argued to death. I think the point is that people in Grandmasters can't really be considered bad players. One can argue circles and circles this way, and by doing that all they come up with is that all statistics aren't enough to be used as viable because one tournament someone played bad, because the top top players may not be #1 on ladder, etc etc. Then we really would never have any usable statistics. I think the ladder gives decently viable statistics because those players in the top 100/even in grandmasters know how to play the game well and can represent the top players of the world, even if they themselves aren't some of the top players in the world.
On another note, I don't really think you should put Mondragon on that list. He doesn't really have any results, besides beating ZeeRaX in the tsl, then losing to cruncher. There is no way he has proved himself as a top player.
|
Points only mean how often you've played, which means nothing. Nothing at all. Win ratio means nothing when you've played only 7 games. Terran win percentage is barely greater than the others, but that could just be because there are more terrans, resulting in tvt being the most common mirror matchup.
This post means nothing at all, it's just some statistics that are misleading.
|
On April 26 2011 05:49 Pelican wrote:
The assumption of homogeneity of variances was also met (p=0.430).
Out of curiosity, you ran a regression on the residual term? Like a breuch Pagan or white's test?
How did the main model that u ran look like? Wouldn't you have problems with omitting of variables, leading to autocorrelation?
Edit: I realize you didn't run a regression so nevermind 
|
On April 26 2011 02:14 the p00n wrote:
In tournaments you usually have a bo3 or bo5 format, whereas on the ladder almost all games are bo1 with minimal metagame or knowledge of your opponent. A terran may excel in a bo1 format (for whatever reason), whereas a protoss or zerg may be stronger in a bo1 < bo[x] format?
Why would lack of metagame and lack of knowledge of your opponent favor terran players so drastically? Besides, The top players face each other more frequently than what is the case for lesser players due to them having fewer possible opponents for each mapseach.
|
Calgary26004 Posts
As you can see, terran is once again dominant in every area.
I can't see that. Can you explain it to me?
|
Getting to the top of the ladder mostly means grinding points against many different oponents, maximizing wins and minimizing losses.
My explanation for this is that Terrans are the more stable race, less succeptible to cheese, more likely to make a comback after a blunder or failed cheese because of the mules, wall-ins and the sheer power of their units when doing counterattacks or harrasment(stim).
When facing completely different oponents every match, you cannot really prepare for your oponent or have a build order prepared for a particular style (like pros in tourneys do), so all you need is a stable play in order to statistically win more than you lose.
Therefore I believe that what we are seeing is the fact that Terran is usually safer than the other races.
|
On April 26 2011 06:15 theJob wrote:Show nested quote +On April 26 2011 05:49 Pelican wrote:
The assumption of homogeneity of variances was also met (p=0.430).
Out of curiosity, you ran a regression on the residual term? Like a breuch Pagan or white's test? How did the main model that u ran look like? Wouldn't you have problems with omitting of variables, leading to autocorrelation? Edit: I realize you didn't run a regression so nevermind
Yah, the HOV test was just a Levene's test.
|
|
|
|
|
|
EPT
