OpenRA

zinc wrote: I'm thinking you would need to collect data on players' ability and their amount of experience with Soviets, and that would be difficult to get in practice. And without those details, I don't see that a large amount of *other data* on game results would be reliable.

Yes but then again, what you say applies for both Allies and Soviets. Data farming with high number of games should be quite effective for general idea but it's obviously not bulletproof. Large games with massive playerbase can do this - if your sample size is like 1000000 games, skill differences between players evens out.

Chimpo wrote: Well the beauty of stats is that you get enough data across different players on different maps then those factors dissolve away

This is only true if you understand and properly account for the biases in the data. If you don't, then the conclusions you draw will sit somewhere on the spectrum between meaningless and wrong.