Author: Andreas Mader
Date: 23:00:43 10/29/97
Go up one level in this thread
On October 29, 1997 at 10:46:10, Robert Hyatt wrote: >I was looking at the official rules to decide how to handle tie-breaks >in the crosstable program I wrote. And found an amusing problem that >goes to show how little thought goes into some decisions that are made >for these tournaments: > >from the rules: > >the first tie-break is the sum-of-opponent's scores. Ok so far. > >the second tie-break, should the first tie-break still result in a >tie is to use cumulative round-by-round scores. And that won't be >fair after accelerated pairings are used. IE a weak program gets >paired against weak opposition in the first round and can easily end >up with +2. A strong program can get paired with a stronger opponent >in rounds 1 and two and end up with a zero. At the end the weaker and >stronger opponents are tied, and the cumulative score favors the weaker >opponent because he won earlier rounds. I think the cumulative round >by round score is a pretty shoddy tie-break scheme anyway, as it rewards >early wins over later wins, when the later wins are harder to come by. >And with the accelerated pairings, it is not just shoddy, it is wrong. > >Another complaint for the complaint department... :) First I want to say that I am very satisfied with your "little" program (how many lines of code and which language??). You did a very good job by implementing the "Buchholz" value (I was wron when I said it was "Sonneborn-Berger", but who cares... :-)). The first tie-break should be "Buchholz" aka "sum-of-opponent's scores". The second tie break should be "Sonneborn-Berger", I try to explain this by means of an example.... Program A won against B and C, had a draw against D and lost against E. "Buchhholz" = score (B) + score (C) + score (D) + score (E). "Sonneborn-Berger" = score (B)*1 + score (C)*1 + score (D)*1/2 + score (E)*0 Given the fact that they used accelerated pairings in the first rounds the cumulative score as a tie-break is nonsense. Best wishes Andreas
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.