Author: Chris Whittington
Date: 02:58:13 11/07/97
Go up one level in this thread
On November 07, 1997 at 05:09:52, Andreas Mader wrote: >On November 05, 1997 at 10:35:42, Bas Hamstra wrote: > >>On November 05, 1997 at 09:34:46, Andreas Mader wrote: >>>Hehehe! >>> >>>Maybe my password for CCC will be canceled by Chris for this but I wrote >>>a little program (with many simplifications, too) and 'played' 100.000 >>>times a 11 round tournament with 33 equally strong programs and one with >>>a 60 per cent win probability. In the end the 'stronger' program has the >>>most points in 28.5 per cent of all the tournaments (many times together >>>with at least one other program). >>> >>>What I have to do now is to evaluate all 34 programs with certain >>>probabilities and write a swiss tournament manager to make this >>>simulation even better and to please Chris.... :) >> >>Great info, please continue! If you could incorporate a rough estimate >>of the relative strength of the programs and the usual way of pairing >>programs I would be very interested in the results. >> >> >>Regards, >>Bas Hamstra. > >I continued playing 'Andreas in wonderland' and now I can enter the >following parameters in my little program: > >- Number of rounds >- Number of participants >- Probability for winning of every program against every other program >- Overall Draw Probability > >The program plays hundrets or thousands of virtual tournaments and >considers the points and the sum of the opponent's points after every >round and at the end of a tournament. > >The program does NOT consider the color (that means that the probabilty >to win is equal, no matter if a program plays white or black) and it >plays no real swiss tournament (that means that the pairing is always >1st-2nd, 3rd-4th and so on, no matter if two participants had already >played a match). Because this program is only made to give a few >statistical numbers I think this doesn't matter too much. > >Now you can >- call me 'Andreas in wonderland' again (Chris Whittington) :) or Ok, but it's a compliment. You can find interesting stuff there :) >- ask questions about certain probabilities (all others). Why not include Swiss into the algorithm ? Or if not: Take the WMCCC results. Make an ELO grading of all the programs from the relative results they have against each other. Use this as your win/lose probability. Use the draw chances as of WMCCC Run Andreas in Wonderland and tell us how the results compare to the 'actual' result. ie. how much is the 'actual' result just a random occurence within the large set of results ..... ? How does this relate to number of rounds ..... ? Chris through the looking glass > >Best wishes >Andreas > >PS.: For a beginning I can take the SSDF list and calculate the >probabilities for winning on base of the current ELO rating or the games >between two opponents or or or.... :) >Maybe next week, too much 'real work' now!
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.