Author: Jari Huikari
Date: 02:22:26 10/04/98
Go up one level in this thread
On October 03, 1998 at 14:10:19, Jari Huikari wrote: >On October 03, 1998 at 07:39:57, blass uri wrote: >>what is the probability to win for every side in a random game? >>If you play many games you can give a good estimate. >I think that probability of a draw is VERY high. Say more than 90%... >Because of '50 moves without progress' rule. (And perhaps a lonely king is >more often stalemated than mated.) > >I can try many games. Nero doesn't have yet '50 moves', 'material' and >'3 repetitions' rules, but I can try playing to some maximum number of moves, >after which I consider a game to be drawn if not ended to mate/stalemate. > >I believe that white and black have same winning probability in a random game. I let Nero recently play many random games. I interrupted after whites 200th move those games which didn't end to mate or stalemate. It wasn't so amusing to watch this playing, because 399 moves were made in few seconds... So I left the machine, and after half an hour saw following statistics: White did 65 checkmates, black did 86 mates, there were 81 stalemates and 545 games were interrupted. (total number of games = 777) So about 80% of these games were draws. And probabilities for white(/black) to win seems to be about 10% for each. It would need much bigger sample for more accurate probabilities. If YOU want to test it with your machine, you can download todays version of Nero from http://www.math.jyu.fi/~huikari/NERO4.ZIP (I'd like to have also feedback about Nero's ordinary playing.) For testing random chess probabilities, give him commands 'beep' (puts horrible sounds off), 'rand' (random playing on) and 'auto' (play self). When you exit from program the statistics are shown in its end texts. Jari
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.