Take a look at this position from one of my correspondence games. (I’m playing Black, and yes, I’m losing.)
What’s the best move for Black? Try to figure it out before reading further.
My Move
On move 20, I dismissed the seemingly strong Bh6, pinning the Queen, because I calculated that Rg5 would start a discovery attack that is indefensible and would force a mate in five. Instead, I played the best bad move I could find: Bd6, trying to hold on in a resignable position.
To my surprise, when the game ended, Chess.com’s Game Review told me Bh6 was the best move and Bd6 was a mistake. Was I wrong? Did I miscalculate?
I tested the position by making the suggested move in Game Review. Sure enough, Stockfish immediately recognized the M5 sequence, while my move only left me down 6 points or so; lost but survivable for a little longer.
It’s weird to be told your move is a mistake, recommended a best move, and as soon as you play that move, the analysis tells you that it leads to a forced checkmate.
This happened with Chess.com’s free Game Review at default Stockfish depth. So I imported it into Lichess which produced a similar outcome at the default depth. So I decided to get to the bottom of it and ran a maximum-depth analysis with my Diamond membership,1 This time, Stockfish correctly identified Bh6 as a mistake and Bd6 as a good move (though Qa5 was slightly better).2
So to answer my own question: yes, the free Game Review’s shallow depth can be wrong in edge cases.3
TL;DR: This is a rare example of me outcalculating shallow-depth Stockfish. 😅
- The Diamond membership is overkill. Platinum is the sweet spot. ↩︎
- If you spotted Qa5, well done! ↩︎
- If we factor in psychology, Bh6 may have still been the better move. My opponent might have blundered by panic-taking my Queen or at least improved my position a bit by taking my Bishop. But psychologically, it’s tough for me to willingly walk into a mate; even if my opponent might miss it. I also try to avoid hope chess. ↩︎