Lichess ratings comparison guide to FIDE, UCF, and Chess.com through statistical analysis • page 1/1 • General Chess Discussion • lichess.org

So chessgoals(.)com compiled a ratings comparison that was updated in January of this year based on ratings of players across Lichess, chess(.)com, and USF and FIDE.

They have a nice chart and everything, but they generally found that chess(.)com and lichess both have inaccurate ratings, but Lichess had inflated ratings (around 200-300 points higher, with ratings inflated up to +/-500 points at the lowest levels) until around 2400 (where it was only 100-200 points higher) while chess(.)com had ratings that were both inflated and deflated compared to FIDE and UCF ratings.

For chess(.)com, on lower levels, players were around 200 points lower than a person's actual FIDE or UCF rating until reaching 2000, and then the ratings were inflated for those 2000 and above, once you get to 2400 on chess(.)com your rating is inflated by at least 200 points.

They put this all into a convenient chart so one can look up their rating and see where they'd roughly place at in OTB tournaments.

https://chessgoals.com/rating-comparison/

So chessgoals(.)com compiled a ratings comparison that was updated in January of this year based on ratings of players across Lichess, chess(.)com, and USF and FIDE. They have a nice chart and everything, but they generally found that chess(.)com and lichess both have inaccurate ratings, but Lichess had inflated ratings (around 200-300 points higher, with ratings inflated up to +/-500 points at the lowest levels) until around 2400 (where it was only 100-200 points higher) while chess(.)com had ratings that were both inflated and deflated compared to FIDE and UCF ratings. For chess(.)com, on lower levels, players were around 200 points lower than a person's actual FIDE or UCF rating until reaching 2000, and then the ratings were inflated for those 2000 and above, once you get to 2400 on chess(.)com your rating is inflated by at least 200 points. They put this all into a convenient chart so one can look up their rating and see where they'd roughly place at in OTB tournaments. https://chessgoals.com/rating-comparison/

NM MrPushwood

#2

You don't know where you'll be in a OTB tourney until you play in one.

what?

I'm 1500 chess.com, and by the chart, it says I'm about 1500 lichess. I'm 1800 lichess, and I don't think any rating of mine is wrong!

AlexiHarvey

edited

#5

When I first started to play on-line I played on chessdotcom. After 1000s of games I came to the conclusion that chessdotcom's rating system at lower levels was completely broke - this was based on subjective experience only. I was also suspicious that such a feature would tend to encourage subscriptions. I no longer play on chessdotcom. I don't play much on LiChess either, however this is due to a preference for playing Bots and dedicated computers - the gameplay is far more consistent and enjoyable - there is also a huge convenience factor.

I also came to the conclusion that you will never get any true sense of improvement playing on-line unless you play a large number of games per day - due to real-life time constraints that's something I can't do, indeed few will be able to do. Simple fact - or belief - is that taking a very small sample from a very large population causes too much variation to be of use for gauging typically very small weekly improvements in player gameplay, especially when you consider player's variation in gameplay is also likely to be huge at lower levels.

The rating system I do use is the one provided by LucasChess which is anchored relative to various chess engines, rather than real-life players. This removes one major source of variation. For what's it's worth this evaluation isn't that much different from my LiChess Rapid rating (measured across 100s of 'computer' games) - as I have suggested there is something broken about chessdotcom ratings at lower levels.

When I first started to play on-line I played on chessdotcom. After 1000s of games I came to the conclusion that chessdotcom's rating system at lower levels was completely broke - this was based on subjective experience only. I was also suspicious that such a feature would tend to encourage subscriptions. I no longer play on chessdotcom. I don't play much on LiChess either, however this is due to a preference for playing Bots and dedicated computers - the gameplay is far more consistent and enjoyable - there is also a huge convenience factor. I also came to the conclusion that you will never get any true sense of improvement playing on-line unless you play a large number of games per day - due to real-life time constraints that's something I can't do, indeed few will be able to do. Simple fact - or belief - is that taking a very small sample from a very large population causes too much variation to be of use for gauging typically very small weekly improvements in player gameplay, especially when you consider player's variation in gameplay is also likely to be huge at lower levels. The rating system I do use is the one provided by LucasChess which is anchored relative to various chess engines, rather than real-life players. This removes one major source of variation. For what's it's worth this evaluation isn't that much different from my LiChess Rapid rating (measured across 100s of 'computer' games) - as I have suggested there is something broken about chessdotcom ratings at lower levels.