Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: New SSDF rating list

Author: Andreas Mader

Date: 22:39:23 12/01/97

Go up one level in this thread


On December 01, 1997 at 21:27:10, Robert Hyatt wrote:

>On December 01, 1997 at 18:13:41, Uwe Immel wrote:
>
>>How could it be, that the possible tolerance (+) for rebel9 is 42
>>points, although he had 10 games more played than hiarcs 6?
>>The more games a program played, the lower should be the tolerance,
>>so it should be just the other way.
>>
>>> SSDF RATING LIST 1997-11-30   57982 games played by  176 computers
>>                                      Rating  +   -  Games   Won  Oppo
>>                                     ------  --  --- -----   ---  ----
>>> 1 Hiarcs 6.0 Pentium MMX 200 MHz    2544   41   -39   327   66%  2424
>>> 2 Rebel 9.0 Pentium MMX 200 MHz     2543   42   -39   337   72%  2375
>
>can also be caused by the ratings of your opponents.  If you play a
>wide range of ratings, your variability will be higher than if you
>play a narrower range of ratings for your opponents...  basic sampling
>theory likes the standard deviation of the population to be small.  If
>it is large, it produces more variability in the estimated result as
>well...

This is true, but it also depends on the number of draws a program has
played. If there are many draws the tolerance will be smaller.

Andreas



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.