Author: Uri Blass
Date: 21:30:51 10/30/05
Go up one level in this thread
On October 30, 2005 at 13:14:28, Graham Banks wrote: >So I don't see any point in running 30 game matches in any future Knockout >tournaments I may decide to run when 20 game matches are likely to produce the >same results. I can add that you can also to choose to continue to 30 games only in part of the cases. If there is a big gap after 20 games then you can be almost sure that the winner is going to be the same. If the gap is small you can continue the match. if I start equal match from result of 11-5 the winner after 30 games is the same in 96% of the cases based on my montacarlo code even if the opponents are at the same level(2.35% of the matches end in 15-15 and 1.65% of the matches end in opposite winner). If you start from result of 10.5-3.5 the winner after 30 games is the same in 97.78% of the cases. I did not calculate it for 10.5-4.5 because the result may be different if the leader is white in the next game or black in the next game but I guess that you can safely say that the winner is the same in more than 95% of the cases when you see result of 10.5-4.5 so you can decide safely to stop the match after less than 30 games if you see that one side has more than 10 points and that the same side is leading by at least 6 points. For 10.5-5.5 the winner is the same in less than 95% of the cases and you have opposite winner in 4% of the cases and draw that half of them will become opposite winner in 3.61% of the cases(of course my assumptions do not hold in your games because the opponent are not equal opponents so practically it may be more than 95% even in case of 10.5-5.5). Uri
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.