Quote Originally Posted by Eldariel View Post
Yeah, ultimately this would crash to player skill. For example, sample would tell you that Anivia loses to Kassadin on lane. However, in a case of few Anivia-players, this is suddenly inversed since Kassadin cannot deal with an AD-specced runepage on lane.

This is not the only lane that hinges entirely on player skill so any sample would most likely be polluted by people misplaying certain lanes (especially accounting for junglers) even on tournament level.
No ... all of those are just statistical variables. And it's still no where near the point.

Seriously. I ask questions - in that post. Those questions are the point. All this nonsense about statistics and sample sizes is not.

But never mind. Lets stick with the statistics then: There is no magic involved in player skill that isn't accounted for by statistics. If champ X beats Champ Y nine times out of ten - then that is the statistic. It doesn't fail to account for the last time where Y beat X.

If the calculations are correct, the sample size sufficent, and the research sensible, the predictions will be statistically correct.