Statistics of lichess ratings (voluntary participation)

Hello everyone! I have a keen interest in both statistics and chess and I wanted to do some statistical analysis of players ratings between Lichess, chess.com USCF and FIDE. I am asking for as much participation as possible. I will compare rating differences between all four using multiple linear regression with SAS.

To collect the data I will be asking for your chess.com username, lichess username, FIDE ID, and USCF ID. In order to participate your rating must not be provisional at any one of them but it is not required that you have all of the above (just a minimum of two). Using some data collection tools I am able to easily link the above although I will manually have to get the chess.com ratings. Please only link your FIDE or USCF ID if you believe it accurately represents your playing strength (aka you have played recently if the last tournament you played was in 1999 it doesn't make much sense to be included for that rating).

Using the data, assuming enough is collected for a good sample, I can estimate a players FIDE, USCF, Lichess or chess.com rating given a series of any other ratings.

Participation is voluntary! You are welcome to PM me your USCF ID or FIDE ID if you don't want it public. I will not share that portion of the data. You are also free to post it below.

Bump I have received 3 responses thus far.

You can do this for me publicly lol

from what i gathered like 1300 fide is like 2000 here or so it seems ... fide players are on a strict diet of killing games taking online rating points for real though it would be nice to see for a non tournament player to see roughly what their actual chess level is,,, i consider myself like a class player maybe expert level if im really playing well but nobody cares about experts or class level... 2200 is when the real chess starts and every player has their own minefield of tricks and tactics that you must navigate, pet lines, side lines ,,, i feel like online chess is a contest where people who don't formally know math answer math questions against mathematicians to see how they do

sounds cool I'm down

I'm going to make some educated guessing here: I have tested none of this.

In the past, OTB players were/maybe still are separated by geographic boundaries. In the past, it seemed very possible that a 1300 player in a higher population density area would not play at the same skill level - whether higher or lower - than a 1300 player in an area with a very low population density.

I'm going to guess that the skill gap has closed from high pop to low pop somewhat due to the corona virus since almost all chess has happened online for the past 3 months. This means that as players go back to their geographic clubs, they bring back 3 months worth of training in a larger pool of players.

So @chessanalyst I'd think you could get enough data without having to poll the community for it if you really wanted to dive into some interesting stats questions about chess ratings.

Plus - I really wish the community would grab on to the idea that online ratings are more relevant in the current day than the local at the board ratings of the past.

Fide- 0 rating/ unrated.
chess.com 2000 bullet, 2000 blitz (inactive since 2 years ago)
lichess- 2300 bullet , 2200 blitz.(active)

2171 US (14636284) 1906 FIDE and on the other bad site I’m dpnorman

@Darksouls

No, 1300 FIDE is NOT 2000 here. Were you exaggerating for a particular reason? If you have accounts on multiple sites (I have 4) you'll see that you're 100 points higher on some sites than others. Lichess ratings are inflated but not by several hundred points.

One poster talked about chess.com as if it were a more accurate rating site than lichess.org. I just don't think that's correct because if you look at the top end on chess.com Hikaru is around 3250 in blitz and last I checked that's a higher number than Magnus's here. Judging from the ratings of elites on other sites, I can only come to the conclusion that ALL sites have inflated ratings but it seems particularly bad at the high end of the distribution.