I was playing a game on lichess, and afterwards asked for a computer analysis, where instead of my perfectly fine move XX the computer insisted on another better move YY, why is that, how can I understand why this another move is better and why I should not play the move I have originally chosen?

I don't see the evaluation score, just the message "Best move is YY", how can I get evaluations and compare my move with the proposed one?

I have tried other chess programs and some of them have recommended my move XX as the best, so it cannot be a total mistake, still lichess analysis does not agree. What should I do and where should I look to understand what's going on?

  • 5
    You should give a bit more detail. Even if you want to keep the question general (i.e. not asking about a specific position), you need to specify at least the engine evaluation before and after the respective moves. For instance you could have missed a win (if the evaluation after XX was roughly equal while after YY it was not), or it could be that you just chose a non-perfect way to win (e.g. if your evaluation dropped after XX, but you are still winning). The latter is very common when you change into a clearly won endgame instead of playing more complicated error-prone positions. Commented Jun 26, 2018 at 12:40
  • @user1583209 I assume the person asking question is familiar with chess, knows about non-perfect ways to win and position simplifications with the slight loss of the material. This question is mostly about "why lichess analysis can differ from other chess engines", because it sometimes IS different and people don't have enough information to figure out WHY.
    – lenik
    Commented Jun 28, 2018 at 5:31

1 Answer 1


TL;DR: Unless your move was marked as "Mistake" or "Blunder" with the significant material loss, you should not worry much about that.

I don't mean you have to ignore everything that computer says, but there are different ways to play chess, and your way may be not as bad as computer thinks. Here are a few points you might want to keep in mind.

  1. The computer analysis is a very time and resource consuming process. What you get on lichess is a bit conservative ("Contempt" parameter is quite low), quick (limited time) assessment of your game, that's supposed to shed the light on the moves where you have lost 0.5 pawn or more. It's deliberately tuned down to avoid risky variants you might be trying to play -- hence the difference in position evaluation.

  2. "We need to go deeper...", once you've got a questionable move as better alternative, you may want to analyse this particular position at full depth. Assuming you're already in analysis window, choose the particular position and move slider to start the analysis on the local computer (see the pic.)

enter image description here

Once the computer analysis has started, you'll see the number of PV lines in the upper right corner about the game notation (PV lines are NOT visible when you just scroll through your game, you have to turn the analysis ON).

Two most important pieces of information you get here are position scores and moves beyond the one played/proposed. The score difference in this case is negligible (less than 0.1 pawn), making top 3 moves approximately equally good, with one of them slightly better than others. Let the analysis run, it usually takes a minute or more, see how the top lines are changing their position and scores.

Once analysis reaches depth 22, it stops. This is considered "deep enough for most purposes". If you want to go deeper, click blue + sign next to the "Depth 22/22". Going beyond depth 22 takes quite a time, try to used that on a relatively fast PC, not on your tables or phone, unless you want to wait for hours.

Finally, here are some common patterns:

  1. Move XX might be not better or worse than move YY, with the deeper analysis you'll see the scores getting equal.
  2. Move YY might appear in the variations as the one of the next moves, being an answer to the threat that's not really visible right now.
  3. Move XX and YY have approximately equal evaluation and constantly change places in the best variation

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