Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: SSDF(Junior 5 - Fritz 5.32)AMD K6-2 450, 1-2, ended 21-21.

Author: Tony Hedlund

Date: 03:45:16 01/06/00

Go up one level in this thread


On January 05, 2000 at 21:59:22, Tina Long wrote:

>On January 05, 2000 at 13:39:25, José de Jesús García Ruvalcaba wrote:
>
>>Hi Tony!
>>	I wonder why this match is 42 games long instead of the usual 40.
>>José.

I had played 39 games before the last overnight session. In that session I got
three games. There's no point in deleting two of them is there?

>partial copy of November list Rebel-Tigger Results:
>Hiar732 K6450     26-16     Fritz532 P200   30.5-13.5

See the above answer.

>I wonder this too, playing odd numbers of games does to a very very minor extent
>skew the results,  ie if every (or say the top 20) program played every program
>in the same number of games there would be one less bit of bias in the results.
>
>I used to use the SSDF tables to make "pretend" tournament tables of the
>computers/programs I owned or was interested in, all play all 40 games, by
>extrapolating ssdf results.  There was always holes, or Genius 4 was leading
>Genius 3 3-1 giving me a 30-10 result, or Genius 3 P90 loses to Genius 3 486/50
>9-11, always spoiling the logic of the tables.
>
>The major finding was that although a program may have been ranked 1 on SSDF, it
>would often be 3rd or 4th when only it's games against top competition were
>considerred.  The extent to which a top program flogs a minor progam can effect
>it's postion amongst the top few too much in my opinion.
>
>eg another partial copy of November list Rebel-Tigger Results:
>MCPro 6.0 P90     17-3     Fritz 3.0 P90   25.5-2.5

Against weaker opponents we usually play 20 games matches. On the Fritz 3 result
the tester might have been away during the weekend having a session going.

Tony

>Don't mind me, I'm just raving here.....



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.