Author: Uwe Immel
Date: 12:45:11 12/11/97
How much do 10 Rebel games with amputated book and
possibly reverse result influence its rating?
In example, based on data below, it should be as following:
------------------------------------------------------------------------
THE SSDF RATING LIST 1997-10-24 57180 games played by 173 computers
Rating + - Games Won Oppo
------ --- --- ----- --- ----
3 Rebel 8.0 Pentium MMX 200 MHz 2506 50 -47 223 65% 2393
Results:
3 Rebel 8.0 Pentium MMX 200 MHz, 2506
Hiarcs6 P200X 9-11 MCP 6 P200MMX 10-10 Rebel 8.0 P90
2.5-2.5
Hiarcs 6 P90 11.5-8.5 MCPro 6.0 P90 12-10 Genius 5 P90
20-10
Rebel 7.0 P90 16-6 Junior 4 P90 6.5-3.5 Geniu4 486/66
15-7
Fritz3 486/66 17.5-4.5 Comet32 P90 26-4
------------------------------------------------------------------------
Result as above is 146 points of 223 games, but if there are only 10
games were won instead of loss (meaning 156 points of 223 games),
the rating of rebel would be *2541*
Of course, that isn’t an accurate rating, because it’s relevant against
which opponent these 10 games are won instead of loss.
I just want to point out, if there is a relative less number of games
played, only a few games with reverse result (in this case just 4,5 %)
will have big influence of the rating.
I don’t know, how much games already played with rebel 9, but imo
the concerned games should be repeat.
Not only rebels rating is inaccurate, the concerned opponents rating too
!
Uwe
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.