EPTNow you release the game to the general public. Each player picks the color they want to be. You notice that more people are playing black 0%, but that the top 200 is 65% black and 35% white. This gets you thinking that black is actually overpowered, except that you don't know that on average, player that play black are about 70 points more skilled than players that play white. Since you don't know this information you try to extrapolate it, except, that you invented a matchmaking algorithm where each player plays games in such a way that their win percentage is always around 50% and at the latest tournament a white player won.
So what information can we gather? The answer? Nothing, it is impossible to know that white is actually overpowered. The information mathematically that you need to determine this is 1. That players who play black are on average 70 points more skilled 2. Two equally skilled players in a black/white game will favor white 55% of the time.
Is there a way to determine 2? Yes, you can test matchups of evenly skilled players who are forced to learn to play as white and as black.
Is there a way to determine 1? No, The rating system being used is a combination of player skill and environment. If the environment is unfavorable this will be reflected as rating. Can we approximate 1? Possibly, if you make a crazy assumption such as that the skill of a player is 95% skill and 5% color. Is there a systematic way of determining this ratio? Yes, if a group of players learned how to play both black and white equally, then you can exprapolate this information.
Why did i post this? Because 1 and 2 are impossible to gather in sc2. It is possible that zerg is actually OP, maybe protoss is OP, unless we have a lot of recent information from similarly skilled players (who have enough games as each race) otherwise we cannot determine either 1 or 2.
When blizzard posts information about win/loss ratio of different matchups what does it mean? It doesn't mean anything. Its possible that player who play terran are simply more skilled on average than zerg, it's possible that zerg adds 50 points to a players skill, but the skill of the players playing it on average lowers it. Maybe cool is an example where he is the same skill as a terran, but that playing zerg gives him an advantage.
The answer is without equal parameters or conditions, all statistics are meaningless.
If you come up with a way for blizzard to get the information about balance without all of the neccesary inforamtion that i just described, please post it and then write a thesis and publish it please.
Also please use probability theory to explain your ideas or at least talk about the minimum information needed.
. I agree. Blizzard has a very very hard job
