Author: Robert Hyatt
Date: 10:26:21 03/13/03
Go up one level in this thread
On March 13, 2003 at 08:43:22, Ricardo R Santana wrote: >I would like to know if there are positions or test sets which I can use to >evaluate rating . > >I would expect test sets which evaluate tatics, endgame, strategy, etc... > > >Thanks >Ricardo Sant Ana It simply is not possible to do this. The approach has to be: generate a set of positions. Try them on a _large_ population of programs. Then fit some sort of curve thru the results based on information known about the ratings of the programs used to run the test. Now you have a curve that might predict the performance of other programs on the same test set. Unfortunately, there are not that many different programs. You need thousands of programs (not different versions of the same program) to calibrate such a predictor, and there simply are not enough (yet).
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.