Author: blass uri
Date: 06:56:37 06/14/00
Go up one level in this thread
On June 14, 2000 at 09:42:06, Torstein Hall wrote: >On June 14, 2000 at 05:47:59, blass uri wrote: > >>On June 14, 2000 at 05:19:58, pete wrote: >> >>>> Program Elo + - Games Score Av.Op. Draws >>>> >>>> 1 crafty : 2035 153 152 20 60.0 % 1965 20.0 % >>>> 2 anmon : 1965 152 153 20 40.0 % 2035 20.0 % >>>> >>>>Individual statistics: >>>> >>>>(1) crafty : 20 (+ 10,= 4,- 6), 60.0 % >>>> >>>>anmon : 20 (+ 10,= 4,- 6), 60.0 % >>>> >>>> >>>>(2) anmon : 20 (+ 6,= 4,- 10), 40.0 % >>>> >>>>crafty : 20 (+ 6,= 4,- 10), 40.0 % >>>> >>>> >>>>According to the "whoisbetter" prgram by Steve Maughan. >>>>From the results of the games "whoisbetter" is 76% sure that crafty is better >>>>than anmon. ........oh well ;) >>> >>> >>>I guess "whoisbetter" is based on common statistics so you would have to >>>continue until you get to 95% to have a significant result but my statistics is >>>really rusty ;) >> >>All this statistics is wrong because results of chess games are not independent >>events. >> >>Uri > >Why do you say that each game is not an independent event? I guess you do not >think that the Comp. on the loosing side is getting psyched out... :-) > >Torstein I think that programs can learn from previous games so the results are not independent. If the programs have no learning function than the results are independent. Uri
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.