Strange Position and Strange Behaviour

[Dicaste]

[d]r3k2r/1pbb1ppp/p7/3pp3/3nn3/P5BP/RP2PPP1/1N1K1B1R b kq - 0 1

Stockfish dev 20250513 Can't find Ba4. Move was obvious to me, I tried Lc0, Torch it finds easily.

[Jouni]

Every move wins:

Analysis by Stockfish dev-20250519-347e328f:

1. -+ (-8.58): 1...Nf5 2.Nc3 Nfxg3 3.fxg3 Nf2+ 4.Ke1 Nxh1 5.Nxd5 Bd8 6.e4 Nxg3 7.b4 0-0 8.Bd3 Bh4 9.b5 axb5 10.Kd1 Rfd8 11.Kc2 Bc6 12.Kb1 Nxe4 13.Bxe4 Bxd5 14.Bxd5 Rxd5 15.Rb2 Rad8 16.g3 f6 17.gxh4 Rd1+ 18.Kc2
2. -+ (-8.54): 1...Ba4+ 2.Kc1 Nxg3 3.fxg3 Bb3 4.Ra1 Nc2 5.e4 Nxa1 6.Bd3 dxe4 7.Bxe4 Be6 8.b4 0-0-0 9.Kb2 Nb3 10.g4 Nd2 11.Nxd2 Rxd2+ 12.Kc1 Rd4 13.Re1 Rhd8 14.Kb2 g6 15.h4 Bxg4 16.h5 Bxh5 17.Rc1 Kb8 18.Rc6 bxc6 19.Bxc6 R4d6 20.Be4 f6
3. -+ (-8.41): 1...Be6 2.b4 Nf5 3.e3 Nfxg3 4.fxg3 Nxg3 5.Rg1 Bb6 6.Bd3 e4 7.Bc2 Nf5 8.Re1 Nxe3+ 9.Rxe3 Bxe3 10.a4 Ke7 11.Ra3 Bf4 12.Nc3 Rhc8 13.a5 Rc4 14.b5 Rac8 15.Ne2 Bd6 16.bxa6 bxa6 17.Ra2 R8c7 18.Rb2 e3
4. -+ (-8.21): 1...Ba5 2.Nc3 Bxc3 3.bxc3 Nxc3+ 4.Ke1 Nxa2 5.Bxe5 Ne6 6.g3 f6 7.Ba1 d4 8.Bg2 Bc6 9.f3 Rd8 10.h4 Kf7 11.Kf2 Rhe8 12.h5 Nc3 13.h6 g5 14.Re1 Nc5 15.Bf1 Kg8
5. -+ (-8.05): 1...Nxg3 2.fxg3 Nf5 3.e3 Nxg3 4.Rg1 Bb6 5.Bd3 e4 6.Bc2 Nf5 7.Bb3 Nxe3+ 8.Ke1 Bc6 9.Ra1 Rd8 10.g4 Nc2+ 11.Bxc2 Bxg1 12.Nc3 h5 13.Ke2 Bc5 14.g5 Be7 15.Rd1 Bxg5 16.Nxd5 Rxd5 17.Rxd5 Bxd5
6. -+ (-8.02): 1...Rd8 2.e3 Nf5 3.b4 Nexg3 4.fxg3 Nxg3 5.Rg1 Bb6 6.a4 Bxe3 7.Ra3 Bxg1 8.Rxg3 Bh2 9.Ra3 e4 10.g3 h5 11.a5 Rh6 12.b5 Rf6 13.Nd2 Kf8 14.Rc3 h4 15.Rc7 Bxg3

[Chris Formula]

Stockfish Tinapa 3.6.4 SE

-9.80/34 1...Ba4+ 2.Kc1 Nb3+ 3.Kc2 d4 4.e3 Rc8 5.Bb5+ axb5 6.Kd3 Nbc5+ 7.Ke2 d3+ 8.Kf1 Bb3 9.Ra1 Bc2 10.Nc3 Nxc3 11.bxc3 Nb3 12.Re1 h5 13.f3 h4 14.Bh2 d2 15.Kf2 dxe1=Q+ 16.Rxe1 Bd6 17.Kg1 Rc6 18.Rf1

-8.60/34 1...Nf5 2.Nc3 Nfxg3 3.fxg3 Nf2+ 4.Ke1 Nxh1 5.Nxd5 Bd8 6.e4 Nxg3 7.Ra1 Be6 8.Rd1 Rc8 9.Kf2 Nxe4+ 10.Kg1 O-O 11.Be2 Ng3 12.Kf2 Nf5 13.Kg1 Nd4 14.Bg4 f5 15.Bf3 Rf7 16.Kh1 Nxf3 17.gxf3 Bh4 18.f4 e4 19.Ne3 Be7 20.Kh2 Rd8 21.Rxd8+ Bxd8

[Dann Corbit]

When everything wins easily and there are no clear and proven checkmates, I would argue that "best move" is nebulous.
I have statistics that show, if there is no forced stalemate by a fortress or eternal check, then +4.5 pawns almost always wins (the non-wins are vanishingly small, just a few instances out of many millions of games).