Please have a look on this game I just played:
lichess.org/gSfCWn0T/Skip to move 27. after I played the move Nf6+. Stockfish notated this move as an incredible blunder: Nf6+??. Also in the game analysis graph this move is notated as huge blunder. But when I turned on the local engine and let it calculate to depth 26 it suddenly realizes this move is still completely winning.
In the right top corner next to "Stockfish 10" you have the "+" button. Click on it and you'll get the better evaluation.
Which won't change the graph nor the notation. This game is just a example how utterly misleading engines can be in some cases.
@ChessAddict1337 After 10 sec it showed +3,9 to me.
You issue is related to your CPU. Maybe you didn't play the best move but there's no doubt that it wasn't winning.
@Proba123 sorry, but there is certainly no issue with my CPU and actually I have no clue what you're talking about. I didn't say anything about any mysterious issue, sorry mate.
Just looking at it, it looks like its throwing away the win, after Bxf6, but he blundered right after that with Nxf6, so u where still completely winning.
Nope, even Bxf6 is completely lost. Like I mentioned before, let Stockfish calculate at least up to depth 26 and it will change it's mind about Nf6+. This move is still completely winning in every variation.
Yes, I agree now.
This is called horizon effect: Stockfish went to 22 plys when the line needs to go to 26 to see the full outcome. I imagine Lichess caps it to make analysis not take forever: for 98% of moves, depth 22 is plenty.
HOWEVER...is it really wrong to call your move, if not a blunder, then a mistake at least? You had two crushing moves available, three other winning moves too, and the move you played required finding ***13 computer-level moves*** in a row to maintain an advantage. You call it misleading: it's correctly told you that you had five better moves...you're going for a *very* technical definition of the word misleading in my opinion.
