
18730 Hits
Something I don't quite manage to get my head around: If the games on the lower level are more random, i.e. skill is in a weird way less of a factor, does the 1200-1600 elo spectrum cover a bigger or a smaller range of "actual skill"?It's supposed to be the "same range" (but that's kind of a weird concept). Elo is purely based off probabilities, if 1200 players win "significantly more" often against 1600 than 1800 vs 2200, I don't see why that wouldn't be reflected in the rankings (so the 1200 player moves up until his Elo-calculated probability reflects his true chances). I guess you are saying something like lower ranked players have a wider variance in performance? Regardless, I think that given enough games Elo should converge to your "true" strength. Now I'm interested in seeing if there is a proof for that.