Author: Robert Hyatt
Date: 21:27:31 09/05/98
Go up one level in this thread
On September 05, 1998 at 16:51:49, Thorsten Czub wrote: >On September 05, 1998 at 08:55:15, Torsten Schoop wrote: > >>Hi all, >> >>the winner is: Chess Tiger! Congratulations! > >I thought this would happen !! Just for discussion, one key point. The Bucholtz tiebreak score is really no good here. Why? Because the initial "seeding" really had no sanity checks done, because there was no good way to do them. The reason this tie-break works in normal Swiss events is that there is a fairly accurate seeding, based on established ratings that are pretty accurate. In this event, this wasn't done because no ratings are available. To understand why this fails, do this: run a swiss where you know the two entries that will end up tied for first. But seed one of the 1, and the other at the bottom. After they tie, compute their bucholtz score. the higher seed will have the higher tiebreak, quite naturally, because when pairing, within a group with the same total points, you use seeding as the first ranking... Were I giving the winner in this event, I would simply list all the programs that finished tied. And I'd say this whether Crafty was 1 on tiebreak or number 5. But clearly, here, the tiebreak doesn't do a thing sensible... > >>Final standings: >> >>Place Name Score Buch. Berg. >> 1-5 Chess Tiger 11.5, 8 73.5 51.75 >> Crafty 15.18, 8 70.5 47.75 > >I think we can see that crafty and tiger have a good position in YOUR >AND in my tournament. >This is an important fact. >I don't think this is random! It shows that these programs have made progress. >I will see if comet can come as close to the top too... the concurrence is very >strong, on the other hand... > >> Comet A95, 8 65.5 41.50 >>Til next tournament >>Torsten > >From Thorsten to Torsten : >thanks for your good work. > >Any ideas who could win MY tournament ?! >Still 5 rounds to go...
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.