Single Comment

Roberto, I was reading about the Glicko method the other day. This is an improvement on Elo that takes into consideration each player's activity. As I was reading about it, it seemed to me that it was addressing some of the same concerns as a weighted system is supposed to address. But instead of weighting the point value of games, it was treating the ratings of more active players as more stable than the ratings of less active players. GCR already does this. So consider a player who intially does poorly at a new game then gets it and starts doing a lot better. So long as he actively plays the game with others, his initial games won't count for as much. If they were against the same opponents he continues to play, each new game he wins against them will lessen the effect of his initial losses. If they were against opponents he no longer plays, they will be considered as less stable than scores against players he plays against more often. Furthermore, if his old opponents don't improve as much as he does, his losses against them won't count as much as losses against stronger players. Although a weighted Elo method might be an improvement on Elo, GCR already comes with features that address the concerns that weighting Elo is supposed to meet. So there seems to be less, if any, need for weighting GCR.