Chess ratings

  
Forum


Chess ratings


Back to forum


Thibault de Vassal    (2007-01-21)
Chess ratings

Hello to all.

I have been submitted this case :

"Rating system: Look at this situation: Current rating ot the player: 2200, 10 games, all players with TER 2200, result 9=, 1+ Case 1: Finished games: 01.01-28.02: 9= --> New rating 01.03: 2200 01.03., no other finished games in this period: --> New rating 01.05: 2228 Case 2: Finished games 01.01-28.02: 9=, 1+ --> New rating 01.03: 2200 --> New rating 01.03. 2214 In case 1 the value of the won game was 28, in case two it was 14. IMO the value of a result should not depend from the number of games you finish in a period. The value of a result only should depend on the rating of both players that they have at the start (preferable) or at the finish date of the game and the result. And the rating formula should be like NewRating = LastRating + SumOfAllValuesOfFinishedGamesInThisPeriod. That's the way (idea) IECG is computing the ratings."


This result looks quite normal to me as a player's level may increase as time passes. Rating calculations are done after periods long enough to avoid big differences... I don't know if you're right about IECG rating calculation, FICGS ratings seem to evolve quicker and I think it's best. By the way this system is used by several chess federations.

Best wishes.


Heinz-Georg Lehnhoff    (2007-01-21 15:16:26)
Chess ratings

I have formatted my message (and partly corrected) to be able to read the text better.

Rating system:
Look at this situation: Current rating ot the player: 2200,
10 games, all players with TER 2200, result 9=, 1+

Case 1: Finished games:
01.01-28.02: 9= New rating 01.03: 2200
01.03. 1+, no other finished games in this period: --> New rating 01.05: 2228

Case 2:
Finished games 01.01-28.02: 9=, 1+ --> New rating 01.03: 2214

In case 1 the value of the won game was 28, in case two it was 14.

IMO the value of a result should not depend from the number of games you have finished in a period. The value of a result only should depend on the rating of both players that they have at the start (preferable) or at the finish date of the game and the result. And the rating formula should be like
NewRating = LastRating + SumOfAllValuesOfFinishedGamesInThisPeriod.

That's the way (idea) IECG is computing the ratings.