Author: Dann Corbit
Date: 15:16:40 04/11/05
Go up one level in this thread
On April 11, 2005 at 16:08:21, Uri Blass wrote: >On April 11, 2005 at 14:29:04, Dann Corbit wrote: > >>On April 11, 2005 at 13:58:03, Uri Blass wrote: >> >>>As far as I know the Israeli chess federation bought in the past a program to >>>calculate rating based on their database. >>> >>>They do not have the source of the program(but only the exe file) and today >>>there is a problem because they decided to change the rules but they have no >>>money to pay to the person who wrote the program so the old rules of rating are >>>practically used and not the new rules. >>> >>>Note also that I thought in the past that it may be a good idea to let everybody >>>to download a file with all the data that they have about results of games but I >>>understood that it is impossible because the data is only in some excel file >>>together with other details that they do not want the public to know so they >>>cannot give people to download the excel file. >>> >>>I am interested to know how other chess federations calculate rating and if you >>>have some good idea that I can suggest to the israeli federation. >>> >>>I thought about some ideas but I prefer to hear first from other people. >> >>This is how the USCF calculated ratings at one time (don't know if it is still >>true): >>ftp://cap.connx.com/pub/tournament_software/USCF.C >>ftp://cap.connx.com/pub/tournament_software/USCF.H >>I wrote this implementation and you can use it for any purpose you like. >> >>This is an interesting project by Royal C. Jones to calculate Elo Ratings: >>ftp://cap.connx.com/pub/tournament_software/prog10.cpp >>And here is his permission to use it: >>ftp://cap.connx.com/pub/tournament_software/Re Your chess rating systems.txt >> >>Elostat is a binary to calculate Elo ratings that is popular: >>http://wbec-ridderkerk.nl/html/download/other/elostat_13.zip >> >>This is a very interesting project by Rémi Coulom: >>http://remi.coulom.free.fr/Bayesian-Elo/ >> >>The FICS source code has the Glicko algorithm. >>ftp://cap.connx.com/pub/chess-utils/fics.1.7.4/ >>ftp://cap.connx.com/pub/chess-utils/FICS.DIST/ >>http://sourceforge.net/projects/chesssrv/ >>http://sourceforge.net/projects/chessd/ >> >>Recommended reading: >>http://www.ratingtheory.com/ >>http://www.chessmetrics.com/ > >Thanks for all the data. > >If I understand the program to calculate USCF rating is free source code by you. Yes. >I still did not look at your files. >The main problem as I see is getting the data and not calculating the rating >based on the relevant formula. You just need to read the game headers. If there are no Elo rating tags, you can set it to whatever you want. >As far as I know the Israeli program simply read the data about the results from >a file but I do not know at this moment details about the file and I only >understood that it is a big excel file that they cannot give to download because >it contain personal information. > >How does your program get the data? Passed in as function arguments. >It is possible to give the program all the data manually but I think that it is >not a good idea because there may be some programs to calculate rating based on >the same data and if every program that calculate rating will need to get the >data manually then it is a waste of work. > >It seems to me a better idea to have the data in some accepted format that every >program that calculates rating get and there can be seperate program to collect >the data and many programs to calculate rating based on the data. It should be a simple thing. You need: 1. White and Black ratings (if any) paired with the players. 2. Wins/Losses/Draws for each player. From that data, you can calculate Elo ratings. If no ratings are available, then any Elo start figure will do.
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.