I just played my first few games of Crazyhouse, and I really don't understand it. At one point I had this position.

[Title "White to move"]
[FEN "r2q1rk1/pbb2pp1/n1p1p2p/1p1nP3/2pPN3/P1P3P1/2BN1P1P/R1BQ1RK1 w - - 1 16"]

Black has no extra pieces, while I (White) have an extra pawn.

Stockfish 10+ on Lichess (depth 18), says I have an immense advantage of +11.2, it and recommends Nxc4, but I have no idea what general strategy is behind this move. I had instead chosen Qh5, which it does not even place in the top 3 moves but thinks is fine too after I make the move, and I know what I am doing with this move.

Upon playing out a few lines extending from Nxc4, I get the feeling that the chess engine may have been mistaken about its value. Is it unreliable for Crazyhouse, or is there some underlying strategy behind that move that I am not seeing?

  • 1
    Can you please post the line?
    – SmallChess
    Commented Oct 24, 2020 at 14:36

1 Answer 1


I can not exactly reproduce it on the lichess analysis board, but the move Nxc4 still seems to be among the top 3-4 moves with a similar evaluation as the top moves. The main idea in this position should be to undermine the dark squares f6, g7, h6 on the king side. Nxc4 aligns well with this idea, since it opens up the bishop's attack to h6 and gives an additional pawn in hand, so after a subsequent P@f6 white still has another pawn for the attack.

The Stockfish fork used on lichess is far above human level in crazyhouse, so usually its analysis should be quite reliable, although it of course also is not infallible and might miss moves/ideas or misevaluate resulting positions.

  • Thanks for your answer! I guess I am underestimating the benefit of having extra pawns; it is as if they are almost as valuable as extra knights. As for reproducing it, as far as I can tell the Lichess Stockfish behaves slightly differently depending on one's computer (as it runs in the browser itself). I was wondering about reliability because I noticed that I can far more easily win in non-standard chess variants against supposedly the same level AI.
    – user21820
    Commented Oct 25, 2020 at 1:55

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.