Author: Uri Blass
Date: 15:07:49 02/20/06
Go up one level in this thread
On February 20, 2006 at 10:44:51, Kirill Kryukov wrote: >Hi Uri, > >On February 20, 2006 at 10:26:43, Uri Blass wrote: > >>I think that it means that you may not find strong correlation between engines >>if the evaluation of one engine is twice of the evaluation of the second engine. > >Yes. An engine may multiply the evaluation by 1.05, or by 2, or by 10, and still >play normal chess. So that "Evaluation difference" table only compares the >evaluation actually reported by engines. > >This is why ponder hit is more reliable table - it compares only things that >engines actually do on board. Expected move is much more reliable than >evaluation. I am only afraid that ponder hit statistics may require larger >number of games. > > > >>I think that it may be better to translate number in evaluation to expected >>result and see correlation between expected results. > >How to translate the evaluation into expected result? You mean just to see if it >is plus or minus? Hmm.. If the program play enough games it is possible to see the average result as function of the evaluation. if you take all cases that it has evaluation of 1.24 pawns for itself then it has some average result in it so the average result is function of the evaluation. The main problem with it is that the program may play against stronger opponent or weaker opponent so the number may be misleading even if you have enough cases with evaluation of 1.24 pawns so a possible idea that I can think of is to make long match between programs with equal strength and use the games to decide about the expected result. Another idea is simply to give the program to analyze many games of players with similiar strength(not of itself) ao it has enough cases that the evaluation says +1.24 and calculate the average result in these cases. 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.