In Atomic chess variant Stockfish 5 sometimes plays very weak in the opening and loses in 3 moves.
Example:
lichess.org/TgDGNFPDI consider it a bug, because Stockfish 5 has rating 1700 and must play stronger (you can't defeat Stockfish 5 by Scholar's mate in Standart)
But if Stockfish 5 doesn't make a mistake in the opening, he demonstrates quite decent skill and for amateur player like me it's not easy to beat him.
There's a simpler 2-year-old bug for Multi-Variant Stockfish, even at its top level, and it's still proving much too complex for me although I'm working on it:
github.com/ddugovic/Stockfish/issues/419The Nc6?? blunder is reproducible locally using the settings for level 5, see below. That SF atomic is quite bad at detecting king capture threats is well-known and we have tried numerous patches in the past to improve this, but none of them proved to gain Elo. There unfortunately does not seem to be an easy solution for this.
setoption name Skill Level value 11
setoption name UCI_Variant value atomic
position startpos moves g1f3
go depth 5 movetime 300
info depth 1 seldepth 1 multipv 1 score cp 80 nodes 74 nps 74000 tbhits 0 time 1 pv e7e5
info depth 2 seldepth 2 multipv 1 score cp 87 nodes 181 nps 181000 tbhits 0 time 1 pv e7e5 b2b3
info depth 3 seldepth 4 multipv 1 score cp 162 nodes 357 nps 178500 tbhits 0 time 2 pv e7e6 e2e3 f8b4
info depth 4 seldepth 5 multipv 1 score cp 961 nodes 930 nps 310000 tbhits 0 time 3 pv b8c6 d2d3
info depth 5 seldepth 5 multipv 1 score cp 109 nodes 2584 nps 430666 tbhits 0 time 6 pv b8c6 e2e3 e7e5 f1b5 d7d5
bestmove b8c6 ponder e2e3
