Introducing a universal rating converter for 2024

I believe lichess blitz and chesacom blitz rating are more reliable than lichess classical and chesscom rapid rating

This is legit

Lichess classical 2k but high deviation
Chesscom rapid 17.5
FIDE standard 16 (but high deviation I believe, since this is my initial rating from a 9 round tournament)

This is legit Lichess classical 2k but high deviation Chesscom rapid 17.5 FIDE standard 16 (but high deviation I believe, since this is my initial rating from a 9 round tournament)

wor4

@NoseKnowsAll First of all, thanks for this!

It was mentioned before, there are roughly 10 times more Lichess rapid players than Lichess classical players. That means of course that there are a lot of players (like me) who only play Lichess rapid and cannot make use of your work.

For now we have the rating comparison by smarterchess https://chessgoals.com/rating-comparison which is very good methodological from what I can tell, but its FIDE ratings seem a bit too high in my rating range (Lichess rapid 2160 = FIDE 1905). It would be nice to have another comparison.

@NoseKnowsAll First of all, thanks for this! It was mentioned before, there are roughly 10 times more Lichess rapid players than Lichess classical players. That means of course that there are a lot of players (like me) who only play Lichess rapid and cannot make use of your work. For now we have the rating comparison by smarterchess https://chessgoals.com/rating-comparison which is very good methodological from what I can tell, but its FIDE ratings seem a bit too high in my rating range (Lichess rapid 2160 = FIDE 1905). It would be nice to have another comparison.

Akavall

@dboing said in #81:

Actually concept of quantile loss is not that hard. It is basically a linear loss with two cases. I think this video: https://www.youtube.com/watch?v=f_XfuS8NwQ4 does a much better job explaining it than I could do here.

@dboing said in #81: Actually concept of quantile loss is not that hard. It is basically a linear loss with two cases. I think this video: https://www.youtube.com/watch?v=f_XfuS8NwQ4 does a much better job explaining it than I could do here.

dvg11