Author: martin fierz
Date: 17:16:19 11/09/02
Go up one level in this thread
On November 09, 2002 at 13:59:43, Jorge Pichard wrote: >When Kramnik or Kasparov play against a group of players with an Elo average of >2600 like in the Olympiad their rating can very easy escalate by as much as 200 >points, but if you change the group of players with a higher Elo average such as >2700 then you will notice how hard it is for them to gain more than 75 >performance Elo rating points. This also hold true for lower rated players, if >you get a player with an Elo of 2200 and face him to play against a group with >an average Elo of 1800 it won't be unusual for him to get a performance rating >of 2400 or higher. > >Pichard. the elo system is actually biased *against* strong players when they play clearly weaker opposition. there was an interesting article by statistician jeff sonas on this topic a few weeks ago (published on chessbase.com, and probably elsewhere too). the only reason that you see such high performance ratings is that if the expected result is something like 8/10, and you win 1 game "too much", and get 9/10, your performance gets a larger boost (+120) than if you happen to get 6/10 instead of 5/10 (+70). it gets more extreme toward the 100% mark of course. this kind of statistical fluctuation always occurs, so when strong players play weaker players, a few of them end up with huge performance ratings, but on average, they are losing rating points. aloha martin
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.