I recently switched from chess.com to lichess.org. This was because I enjoy live chess more and it seems like can find more live chess on Lichess while it chess.com keeps you waiting. Chess.com searches for who they think is your perfect opponent whereas Lichess lets you play whoever you want.

However, I’ve noticed that Lichess gives more points to the winner and takes away more points from the losers, and I mean much more. I don't know if I'm correct though, but I think I am. Lichess also gives youn a higher starting rating.

If I'm right, then which online chess website gives the most realistic chess rating to its users?

  • 6
    lichess.org uses the Glicko-2 rating system, which adjusts ratings by a greater amount when it is less sure of your rating, like when you first start playing. It is designed to be more "correct", not less.
    – dfan
    Jan 6, 2016 at 2:17
  • 4
    Do you mean which server rating is closest to Elo or which server rating has the best predictive value? Jan 6, 2016 at 9:34
  • 4
    @BlindKungFuMaster closest to elo
    – Lynob
    Jan 6, 2016 at 9:43
  • 3
    Your question is philosophically meaningless. All chess ratings are somewhat arbitrary and at best tend to capture past performance not what a person is actually capable of doing at this moment in time. Further they each use different systems, on different groups of people so are not at all interchangeable. If you care about elo then you should rephrase with which rating is LEAST WRONG. Feb 25, 2020 at 1:23
  • 3
    There is no definition of a "correct" rating.
    – David
    Sep 7, 2020 at 6:40

6 Answers 6


How do you define realistic for chess ratings? Actually, they are all realistic. Ratings should be considered inside their own population because your rating in server X is your performance relative to other players in the same server. As it depends the population you shouldn't compare one rating to another rating in another server. But if you really want to compare your rating to another population approximately, please see this method: How does Chesscube.com rating compare to USCF and FIDE ratings?

All chess rating systems follow almost same rules, but they have different converging speeds. This is what you explained in your question, one server gives less on win and takes less on loss, another one gives/takes more etc.

There is a thing, K-factor, in USCF rating system for example, to tweak this converging speed from game to game. Please see this topic: What does "half-K" mean in the USCF tournament rating system?

When you enter a rating population as a new player, you start with average rating of that population(generally). But it doesn't matter which rating you start with, even if you start from 3000, you will converge to your real rating as you play games. So, it's not about being realistic again.

And generally for all rating systems, your first 8 to 20 games will have super high converging speeds, to approximate your rating faster. After that, converging speeds drops to normal and ratings get stabilized.

So, shortly, your ratings will be different in different populations, and it's totally normal.


My Lichess blitz rating is very close to my FIDE classical rating. But is that correct?

The FIDE rating is based on slow (six hour), over the board games, in my case usually in team matches, one per match weekend, where I make sure I am well rested and then try to play to the best of my abilities. For instance, if I suspect a trade to a better endgame is available, I will calculate it and go for it.

On the other hand my online blitz games are purely for fun, with a glass of whisky at the end of the day when I am dead tired. I play the first move I think of with major preference for just throwing stuff at his king and flashy sacrifices. I avoid all endgames as they're boring.

It seems more coincedence than correctness.


Well, assuming that you define a 'correct rating' by a FIDE rating, I'd say that chess.com is up there, although not necessarily for the bullet pool. However, for the slow time control pool, chess.com is extremely similar to FIDE and maybe even more similar to USCF. On chess.com, you start at 1200 which is quite similar to an average scholastic rating. Furthermore, lots of cheaters on chess.com deflate the ratings a little bit.

P.s. lichess.org is extremely inflated. 1800 FIDE basically equates to around 2250 on lichess (classical time control). On the other hand, 1800 FIDE equates to around 1850 on chess.com (again, classical time control).

Did I approach your question right?


You should understand what your rating really is: A rough approximation of your recent performance against other players in the same rating pool. What's good on one server might stink on another, and none of it matters.


Like other answers, let's assume FIDE rating is the baseline for comparison. This is reasonable as it's widely understood by chess players; for example it's what titles like FM, IM, and GM are based on.

Then it is very much possible to compare a player's strength across multiple sites and see which matches the baseline the closest. This comes with some assumptions, like assumptions that online chess performance matches OTB performance and performance across time controls, which appear decently reasonable.

A fairly recent new site lets you compare directly with a graph: https://www.chessratingcomparison.com/Graphs

According to https://chessgoals.com/rating-comparison/, which is based on the player data from their database, chess.com does a fairly accurate mapping:

The Chess.com ratings map pretty closely to USCF and FIDE ratings overall. You can use it generally as a quick mapping to your USCF rating. USCF ratings start at 100, and the Chess.com equivalent is slightly under 500. Between 100 and 1750, the 400-point gap shrinks as the ratings go higher. Above 1750 we see Chess.com ratings are higher than OTB on average. FIDE ratings follow a similar distribution to the USCF ratings. Fewer OTB during the pandemic has not impacted Chess.com to over-the-board comparisons.

Typical bullet ratings on Chess.com are slightly lower than blitz ratings. Rapid ratings start higher than blitz ratings, with a breakeven point around 1900 when blitz ratings begin to be higher. This table also accounts for the change that Chess.com made to move 10+0 games to the rapid category. Chess.com to chess.com ratings have about 7000 data points, USCF 1500 data points, and FIDE 1900 data points. See the complete data below.

Update as of Dec 2021: The page also confirms common knowledge that Lichess ratings have gotten very inflated at lower ratings, and the discrepancy gets smaller for higher ratings, since Lichess's rating system starts users off at a very high starting rating of 1500.

We have also analyzed how lichess.org ratings map to OTB ratings. We combined the first two tables to create the lichess vs. OTB comparison. Lichess.org ratings are generally higher than OTB ratings, and the separation increased in 2021. If you have a lichess rapid rating of 1800, don’t expect to score an 1800 performance at your first FIDE event. Your rating will map to roughly 1420 FIDE or 1270 USCF. The complete data set is below, which maps all lichess ratings to their OTB averages across our survey results.

In particular, a Lichess rapid rating of 1800 corresponds to only FIDE 1420 rating, Lichess rapid 2000 is about FIDE 1660, and Lichess rapid 2400 is about FIDE 2300.

The details of their calculations are here https://chessgoals.com/rating-comparison-explained/

To compare Lichess and chess.com, my rule of thumb is Lichess is roughly 300 points higher than chess.com, give or take 100 points. The hand-wavy explanation is that chess.com starts unrated players at 1200 and Lichess starts at 1500.


Opinion and subjective.

There is no way to tell unless you provide a very detailed description of how you would evaluate each site.

And no rating system is that accurate even if all the players were consistent and no beginners were improving to further disrupt things.

Even if it could be accurate no system would have enough games from enough people to be accurate except in small groups that have many common opponents. That is why when a hotshot from moosejaw or podunk goes to the big city to play in a tournament they get creamed by lower rated players because theirs had inadequate range of competition to be valid.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.