Single Comment

Other weakness I see is that You don´t know how many games are needed to consider a rating to be 'somewhat confident'. It is very possible that a player with only a few games played, say less than ten, but with almost perfect score against 'well rated' players, show a rating that does not reflect the player´s force, being the rating, perhaps, much less than other player´s rating with a lot of games played but much less average and relatively worse record against others. It has been said that the rating must stabilize with time, but I´m not sure how many games are needed, and the disparity in number of games may introduce a bias that can give ratings that could be not so easy to compare with accuracy. But once 'stabilized', the whole history introduces another bias, product of very old games considered with the same weight as new ones, this is the main reason I insist with the weighted history idea.