Comparison of lichess game analysis with chess.com analysis

I played this game on lichess, and I got 91% accuracy based on lichess analysis. On chess.com they are saying it should be closer to 75.7.

Link to chess.com thread (only for reference, I am not promoting one site over the other here, and just interested in game analysis regardless where it was played.)

www.chess.com/forum/view/general/does-it-make-sense-you-could-have-2-blunders-and-91-accuracy

Full game with lichess analysis.

LiChess is such a better site. I gave up on chess.com for several reasons. I think it was king for a long time, but LiChess has continued to grow and improve. The learning tools and database and analysis options here have gotten better and better.

That doesn't really answer my question. This is not a chess.com vs. lichess thread. I am asking about the evaluation discrepancy. If you can back up either one regardless of your allegiance to one site, that would be appreciated. It's also plausible they are measuring two different things, like maybe lichess evaluated the opening and middlegame but didn't "deduct points" for the sloppy win in the endgame.

Let's keep the discussion to the game evaluation and not talk about the sites.

Obvious Lichess is Better than any site in Nature of this Site, I feel like it was created to make Chess really like they are, and it's style and abilities , even it's free , telling of that. and the fact it's free will keep it always Best, so We may donate or help on our own will, not like on chess com or some another

Well the evaluation graphs look about the same so it's probably just the specifics of how accuracy is defined. The Lichess way is described here: lichess.org/page/accuracy

@chess2025chess said in #3:
> That doesn't really answer my question. This is not a chess.com vs. lichess thread. I am asking about the evaluation discrepancy. If you can back up either one regardless of your allegiance to one site, that would be appreciated. It's also plausible they are measuring two different things, like maybe lichess evaluated the opening and middlegame but didn't "deduct points" for the sloppy win in the endgame.
>
> Let's keep the discussion to the game evaluation and not talk about the sites.

What if I don't want to? You're gonna cry? You're making an evaluation of an aspect between LiChess and Chess.com. Obviously, I would conclude the evaluation aspect of LiChess is better, but you are wrapped up in your meaningless little chess game. Furthermore, you just want to compare the two sites on your own terms. I think LiChess is the better site. So, there. You're acting like you played some sort of world-class chess game or something. You didn't.

I think lichess analysis is better since it uses stockfish 17

@billcourtney said in #6:
> What if I don't want to? You're gonna cry? You're making an evaluation of an aspect between LiChess and Chess.com. Obviously, I would conclude the evaluation aspect of LiChess is better, but you are wrapped up in your meaningless little chess game. Furthermore, you just want to compare the two sites on your own terms. I think LiChess is the better site. So, there. You're acting like you played some sort of world-class chess game or something. You didn't.

You're a real tearJERKer, you know that?

The analysis that they provide is fundamentally the same, since both use top engines which are very close in strength. The fact that one can be 90% and the other 70% just shows that the methodology for the score is different, and so it's meaningless to compare numbers.

More important is to focus on the analysis itself, the mistakes etc. The major mistakes will be the same for both sites

@chess2025chess said in #3:
> I am asking about the evaluation discrepancy.

For both sites, accuracy is derived from how frequently a player made moves a chess engine disapproves of, however, both sites then adjust that number for different purposes.

Chess.com adjusts the frequency to resemble school grades: support.chess.com/en/articles/8708970-how-is-accuracy-in-analysis

Lichess.org adjusts the frequency to reflect winning chances: lichess.org/page/accuracy

Notice that your blunders happened when you couldn't recognize a forced mate, while your opponent's blunders led to the forced mated sequences you missed. For chess.com, that gets you an "average" grade, while on lichess.org, that gets you better "winning chances" because you were in winning positions.