Database Window - Rating – Create Elo Start List

 

Unix inventor and computer chess pioneer, Ken Thompson, has developed an algorithm which allows one to create an Elo rating list out of an arbitrary set of games. Any database of games can be treated as a gigantic tournament. Each player gets the same initial rating (e.g., 2400). After evaluating the results of all games in the database, each player gets a new rating. Using these new values, the games of the database are rated again. This is done over and over again, until the ratings of all players stabilize and the values remain constant.

 

 

The Elo management in our chess program was implemented mainly in order to evaluate engine tournaments. But it is very interesting to use it on human results as well. It can also be quite exciting to create Elo lists for historical game data. For the system to work properly, it is absolutely critical that the players’ names are completely unified. We recommend using the large high-quality databases from ChessBase (e.g., MegaDatabase) for the creation of historical Elo lists. Most other databases contain player names with different spellings and other inaccuracies. In addition, databases which do not contain complete tournaments will distort the ratings. For instance if they only have the best games of certain players these players will achieve very high ratings from the program.

 

For engine ratings, a good Elo list should be based on at least 300 to 500 games. Implausible values in the start list are not a big problem, because after a few engine tournaments, the programs will approach their real ratings (underrated engines will shoot up, grossly overrated ones will lose point very quickly).

This is how to generate an Elo list

 

   First, select all the games in a database list that should be used for calculation. If you want to use the whole database, press Ctrl-A to select them all.

 

   Click Database– Ratings– Create Elo Start List.

 

   The program prompts you for a file name. If you use an existing Elo list, its contents will be overwritten.

 

   If there are not enough games per player in the database, the program informs you, and does not calculate any ratings.

 

The Thompson algorithm calculates relative playing strengths. It assumes that the average rating of all the players in a rating list is 2400. This will normally not be the case. To get correct practical ratings, it is necessary to rescale or gauge the list by adding or subtracting a certain percentage from each player. The best way to do this is to take a very stable player with a known rating, and add or subtract an offset to make the rating equal his real rating. The program will then adjust the ratings of all other players accordingly.