Author: Dieter Buerssner
Date: 11:33:05 12/08/03
Go up one level in this thread
On December 08, 2003 at 14:12:50, Drexel,Michael wrote:
>Test results Crafty 19.06-Hiarcs 9:
>
>Crafty 19.06 (DC) - Hiarcs 9 88,5 - 161,5 (+44 =89 -117)
> (35,4%) (64,6%) 250 games
>
>Crafty 19.06 (CB) - Hiarcs 9 77,0 - 173,0 (+29 =96 -125)
> (30,8%) (69,2%) 250 games
This still could be explained by random fluctuation and a bit of bad luck. A
Monte Carlo simulation (assuming independent games, of course - no learning
etc.) shows (I made up some number, parameters are for the first match:
88.5-161.5):
Result of chess matches
Player A as white wins 18.8%, draws 38.3% and loses 43.0%
Player A as black wins 16.4%, draws 32.8% and loses 50.8%
Expected result: 35.34% (as white 37.89%, as black 32.79%)
A match of 250 games was simulated 1000000 times by a Monte Carlo method
result probability p <= res. p > res.
[...]
76.5 - 173.5 ( 30.6%): 0.4525% 2.3652% 97.6348%
77.0 - 173.0 ( 30.8%): 0.5323% 2.8975% 97.1025%
^^^^^^^
77.5 - 172.5 ( 31.0%): 0.6123% 3.5098% 96.4902%
[...]
87.0 - 163.0 ( 34.8%): 3.3196% 42.8417% 57.1583%
87.5 - 162.5 ( 35.0%): 3.3460% 46.1877% 53.8123%
88.0 - 162.0 ( 35.2%): 3.3515% 49.5392% 50.4608%
88.5 - 161.5 ( 35.4%): 3.3856% 52.9248% 47.0752%
89.0 - 161.0 ( 35.6%): 3.3451% 56.2699% 43.7301%
89.5 - 160.5 ( 35.8%): 3.3444% 59.6143% 40.3857%
90.0 - 160.0 ( 36.0%): 3.2545% 62.8688% 37.1312%
[...]
Average result of simulation 35.3376%
So still around 3% probability for a result of 77-173 or worse, when the real
strength is assumed 88.5-161.5.
>Crafty 19.06 (DC) - Hiarcs 9 TC 74,5 - 125,5 (+39 =71 -90)
> (37,3%) (62,7%) 200 games
This is a different setup for Hiarcs only?
Regards,
Dieter
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.