A friend told me its very hard to implement a bughouse chess engine because it requires much more processor power than a regular chess engine.

Are there any bughouse engines out there ? if not, what are the reasons ?

  • Among bughouse's added complexities that answers below point out, I think (but I'm totally speculating) that the multi-board aspect is probably a more significant issue than drops per se (of course, the drops are moving from board to board in the case of bughouse, so they're not unrelated). My only reason for saying this is that the best Shogi (which has drops) engines do rival professional human players. en.wikipedia.org/wiki/Computer_shogi#Computers_versus_humans
    – ETD
    Jul 20, 2013 at 22:32
  • 2
    The main reason that bughouse engines are fewer and weaker than normal engines is simply because fewer people play bughouse than normal chess. This means less interest in building an engine for it, because in principle bughouse should be perfect for engines seeing that it is so immensely more tactical than normal chess.
    – Halvard
    Jul 21, 2013 at 5:41
  • It's key to distinguish "one board" engines that play the moves of a board, and with another human or engine make a bughouse team, and a "dual engine" wich controls both boards and thus can make combined board strategied instead of being at mercy of the unpredictable pieces flux the human sends him. Jan 12, 2017 at 1:49
  • a dual engine can sac on one board because it sees a strong drop on the other board, but if it's a human, he may drop the passed piece in a weak spot. Jan 12, 2017 at 1:51

4 Answers 4


There are certainly several chess engines, but as a result of the additional complexities referenced by @TonyEnnis, these engines are weaker than reasonably strong players. It's extremely difficult to take into account the different scenarios that can arise from two positions as pieces switch from board to board every move. The resulting search tree increases at an exponential rate even greater than that of standard chess.

The engine that I'm most familiar with is Sjeng which claims that it has an ELO of around 2000 on FICS, equivalent to a strong human player.

  • Accepting as you pointed to Sjeng
    – Ofiris
    Jul 21, 2013 at 19:36
  • I am guessing 10 years later a team with two engines or one engine playing both boards would be more dominant than standard chess. Human/engine team would be interesting to watch -- i would guess a strongish human (2000 or even 1800) with a top engine might be much stronger than two humans of 2000 strength because the engine might be very good at figuring out how to play on a per-partner basis, maybe even remembering past positions with that same partner.
    – releseabe
    Jul 26, 2023 at 14:39

Sjeng is your best bet but play with an engine is very weird. Typically a human + engine team will have their play revolve around getting up on time (sometimes even by only a few seconds). Then the human player will tell the engine to move fast, and then sit. Note that since the computer can play decent moves instantly, that effectively means it is impossible to sac material against the human player unless also up on time -- even if the sacrifice is very strong and leads to an imminently winning attack or quickly recovering the material. The human player will accept the sacrifice(s) and sit, then let the engine win on time or use the extra material on the board to win. For these reasons none of the strong bughouse players will accept games with engines anymore, really.

  • 1
    very interesting, you seem to know about bughouse, don't doubt about posting more! Jan 12, 2017 at 1:36
  • @Santropedro: This rule-set is fair and yet guarantees progress, and also prevents the kind of issues that causaSui mentions here, because any player with at the obligated board may change the obligated board and hence force the other board to make progress even in the absence of time controls.
    – user21820
    Jan 9, 2022 at 7:08

I don't know of any bughouse engines.

Extra complexities:

In addition to playing the standard game...

  1. Extra tactical vision. The computer now has to envision good moves that either side could execute if only a certain piece were in-hand.
  2. Drops make the search tree enormous. There's really no restriction on where pieces may be dropped (pawns? can pawns be dropped?) That makes the search tree enormous.
  3. Multi-board sacrifices. If the machine were to find a good move "if only I had a [piece] in hand,", it then has to decide if it is wise to sacrifice on the other board (or even this board) to get that piece.
  4. Tactical clock management. The bughouse games I played were always on the clock (that is - 5 minute bughouse.) Tactical clock management is another complexity. I've seen games won when one side had a clear win on one board, and stopped playing on the other boards when they held the time advantage, to starve the opponents of extra pieces.
  • Good answer, thought (2) also happens in crazyhouse, and current engines are strong ( much weaker than chess ones of course). Actually lichess's stockfish crazyhouse engine id the strongest according to recent informal tournaments, but was beaten up by a lichess top 20 crazyhouse human. But, the engine used like 3 seconds of 5 min. with more time may become superhuman. Jan 12, 2017 at 1:44

Sunsetter is best seems, but Sjeng (especially 11) is good for many purposes too.

See http://bughousechess.wz.cz/BughouseEngines/

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