Ok, I'm actually pretty sure, that it does; so the question should be: Does it matter? First, let me explain, what the problem is: When you start a QM, you are matched via the match making system against an a priori unknown opponent. Knownledge about counterpicks helps in that first match only via the current meta (popular characters, weekly free caracters...). I consider this as the fair match. The problem arises, when both players agree to rematch and the loser (let's call him Bob) of the first match changes his character (or the bank in Puzzle Strike). So Bob now has an advantage, because he new the opponent (and bank) and could counterpick. Still the outcome of the rematch effects the Elo score of both players in the same way as the first (fair) match. Consequences: Someone playing to maximise his Elo should never agree to rematch after winning and always try to rematch and counterpick after losing a match. Also, using statistics about top or average Elo scores to estimate character strength/balance becomes unreliable, since some characters are more prone to counterpicking, while others might be used mostly as a counter pick and different players behavior might bias this even further. There are different possible solutions: Don't let rematches after a QM influence the Elo score When rematching after QM, allow both players to change characters Use a seperate Elo score for groups of rematches, such as best of 3/best of 5 (first game effects the character Elo, the whole game (all matches with standard counter picking) effects the best of 3/5 Elo score I know, those solutions aren't very elegant. That's why I ask: Is the problem important enough to even consider the trouble with those solutions?