lelorrain (1553) ☗5☗1☗92018-03-21 17:56

Hello everyone, 

I would like to know if a survey of the rating system has been done and if it is possible to immagine an evolution of the existing formula. 

The existing calculation is really sensitive and generate tension that tends not to play with everybody. I mean, if I have an Elo Score equal to 1500 and if I loose versus a 1200 player, I'm going to loose a huge quantity of Elo, just for one game! And it is going to take many games to compensate. 

As a result I have no interest to play versus unrated players, vs to strong player or lower players... It is a lot of players to avoid.

On the Lichess chess website (one of the top chess players site), this is the reason why they used the Glicko2 system. It is integrating additional parameters. If you loose once versus a lower player, you loose a bit of points. More you loose versus lower player and more you loose points each time. It allows mistakes. It is also considering the % of winning of a player: If a player has 100% of win with a score at 1200 for example, he his certainly stronger than its estimated level. It is taken into account as well.

The glicko system is not an absolute system, but it is integrating rational and interesting parameters. Did you planed to integrate this kind of approach ?

Best regards!