Lichess rating inflation • page 1/1 • Lichess Feedback • lichess.org

Hello friends.

I am interested if anyone else has noticed the strange rating inflation on lichess, compared to the other most used sites (namely chess24 and chess.com). My rating on chess.com is 1500ish, on chess24 1650ish, which is almost consistent with my results in otb blitz tournament play (1578), but strangely my rating on this site is 1900, with regular good results with 1800 guys.

I feel I am having similar challenge when playing someone 1850 on this site like I have when playng 1550 on other sites.

Anyone else noticed this?

Hello friends. I am interested if anyone else has noticed the strange rating inflation on lichess, compared to the other most used sites (namely chess24 and chess.com). My rating on chess.com is 1500ish, on chess24 1650ish, which is almost consistent with my results in otb blitz tournament play (1578), but strangely my rating on this site is 1900, with regular good results with 1800 guys. I feel I am having similar challenge when playing someone 1850 on this site like I have when playng 1550 on other sites. Anyone else noticed this?

CreativeThinking

@banem82 lichess uses glicko-2 rating system.

banem82

Can you be more concrete?

CreativeThinking

@banem82 this may be helpful: https://lichess.org/page/rating-systems

BACollin

@banem82 You can read more about the Glicko-2 rating system at lichess.org/faq, or a more detailed explanation of the calculation at http://www.glicko.net/glicko/glicko2.pdf.

The short answer is that chess.com, chess24, lichess, USCF, and FIDE all use different rating systems. In these different rating systems, the starting rating (1500 on lichess and 1200 on chess.com) are the average ratings of players (roughly). Basically, since lichess has a higher starter rating, and therefore higher average rating than chess.com, it has generally higher ratings. There isn't rating inflation between lichess and chess.com because they have separate calculations, and are therefore incomparable. There IS A LOT of rating inflation on chess.com where the highet bullet rating is 3500.

You can't compare your ratings across all of these platforms as more or less accurate than the other platforms because the calculations are different; each of your ratings on each platform is accurate to the body of players on the platform, and their respective ratings.

Lichess uses the Glicko-2 rating system, chess.com uses the Glicko rating system, chess24 uses a variation of ELO, USCF uses a different variation of ELO, and FIDE uses a different variation of ELO. In general, the rating systems devised by Mark Glickman (Glicko and Glicko-2) are 'more accurate' than those devised by Arpad Elo because Glickman's systems calculate a rating deviation, which measures the accuracy of your rating and influences the calculations. So objectively, Glicko and Glicko-2 are more accurate than ELO, but each of these systems are accurate to itself and cannot be compared.

@banem82 You can read more about the Glicko-2 rating system at lichess.org/faq, or a more detailed explanation of the calculation at http://www.glicko.net/glicko/glicko2.pdf. The short answer is that chess.com, chess24, lichess, USCF, and FIDE all use different rating systems. In these different rating systems, the starting rating (1500 on lichess and 1200 on chess.com) are the average ratings of players (roughly). Basically, since lichess has a higher starter rating, and therefore higher average rating than chess.com, it has generally higher ratings. There isn't rating inflation between lichess and chess.com because they have separate calculations, and are therefore incomparable. There IS A LOT of rating inflation on chess.com where the highet bullet rating is 3500. You can't compare your ratings across all of these platforms as more or less accurate than the other platforms because the calculations are different; each of your ratings on each platform is accurate to the body of players on the platform, and their respective ratings. Lichess uses the Glicko-2 rating system, chess.com uses the Glicko rating system, chess24 uses a variation of ELO, USCF uses a different variation of ELO, and FIDE uses a different variation of ELO. In general, the rating systems devised by Mark Glickman (Glicko and Glicko-2) are 'more accurate' than those devised by Arpad Elo because Glickman's systems calculate a rating deviation, which measures the accuracy of your rating and influences the calculations. So objectively, Glicko and Glicko-2 are more accurate than ELO, but each of these systems are accurate to itself and cannot be compared.

banem82

thx guys, I wasn't aware that these different rating systems produce such different results

Joan_of_Arc1

A chess.com rating of 400 is already like a lichess rating of 1000
a chess.com rating of 1000 is probably already a lichess rating of 1500
the gap decreases as ur rating goes higher
a chess.com rating of 1700 is around a lichess rating of 2000
a chess.com rating of 2500 is around a lichess rating of 2600
and soon, chess.com ratings get higher than lichess
a chess.com rating of 3000 is lichess 2700.
chess.com has some ppl at 3300, and those guys are only like 2900 lichess

A chess.com rating of 400 is already like a lichess rating of 1000 a chess.com rating of 1000 is probably already a lichess rating of 1500 the gap decreases as ur rating goes higher a chess.com rating of 1700 is around a lichess rating of 2000 a chess.com rating of 2500 is around a lichess rating of 2600 and soon, chess.com ratings get higher than lichess a chess.com rating of 3000 is lichess 2700. chess.com has some ppl at 3300, and those guys are only like 2900 lichess