Author: Keith Hyams
Date: 10:50:49 08/10/05
o Program Score % Av.Op. Elo + - Draws
1 Fruit 2.1 159.0 / 270 58.9 2469 2532 37 35 31.1 %
2 Chess Tiger 15.0 15.0 / 30 50.0 2532 2532 109 109 33.3 %
3 The King 3.33 14.0 / 30 46.7 2532 2509 80 130 46.7 %
4 Fritz 8 14.0 / 30 46.7 2532 2509 80 130 46.7 %
5 List 512 13.5 / 30 45.0 2532 2498 110 127 23.3 %
6 Junior 9 13.5 / 30 45.0 2532 2498 110 127 23.3 %
7 Ruffian 2.1.0 12.5 / 30 41.7 2532 2474 125 122 16.7 %
8 Toga II 0.93 10.5 / 30 35.0 2532 2425 91 112 43.3 %
9 Hiarcs 9 9.5 / 30 31.7 2532 2399 131 108 23.3 %
10 Aristarch 4.50 8.5 / 30 28.3 2532 2371 137 104 23.3 %
The chart above has been copied from Kurt Utzinger's website. Kurt has not
objected to my using it. (Apologies to Kurt for poor formatting)
The error bars (headed + and - ) have been produced with Elostat. My feeling is
that, at any given number of games, the possibility of an engine being rated x
points too high should be equal to the possibility of the engine being rated x
points too low. Therefore if there is a possibility that Aristarch is rated 137
points too high, I would expect there to be an equal number for the possible low
rating - At any given sampling point I would expect error bars of the same
length above and below zero.
Would somebody please explain to me how Aristarch can be +137 -104.
Regards
Keith
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.