lichess.org
Donate

Endgame with a lot of pawns and some small figures

Can someone explain with words what Stockfish thinks that black should have done at the 48th move?

My thoughts were: I have no idea how to continue, let's trade the figure plus I have for a pawn, maybe I can do something with the pawn. But stockfish did know how I could've continued, or so it claims, and I don't understand any move in its variation.

Thank you very much. I promise I will read some books about endgames.
It might just be a draw.

Something you should not have done tho is to play h4. The plan with two knights would be to plant one on h4 and the other on g4. Then you could play Nxh3 gxh3 g2 with the h4 knight guarding the g2 pawn. Also h4 gives a potential hole to move your king up the board after this sac. Like Kg5, kh4, kg3. Same goes for your bishop. The h4 pawn push restricts the movement of your pieces for little to no gain. It might even guarantee the draw.

Also, the win, if there was one, would be easier with queens on the board in this instance. Mainly because you could bring your bishop to d8 and have your king guard it. And then white queen is shut out of the game so no perpetual or counterplay. Then you could set up your pieces and queen and prepare a sac on h2 followed by a g3-g2 pawn push.

Computers are terrible at evaluating such closed positions. To a human, your bishop is locked behind your pawns and so it's like not having a bishop. But to a computer it's an extra 3 points. Even if it is useless long term.
Oh I meant to put the N on the g5 and h4. See how they attack the light squares? Another idea is to take with a knight on f3, gxf3, Nxf3. Then bring the knight back to h4 and play f3. Then Bg5 to activate the bishop and prepare g2 and Be3 attacking/protecting the queening square.

None of this is possible though with a pawn on h4.
I had a glance at it and I have to admit I would have done
the same as you, dont know if its correct
but I know as vomere said that this long term planning
for a computer does not seem to be a useless piece
I think It count its nevertheless
I am doing an analysis with a different engine now,
houdini, and it gives black advantage of three points or so
but all it does it is moving the pieces around without breaking the barrier,
so I guess at that point c6 is correct.
So the mistake was previous thats for sure, not on 48.

This topic has been archived and can no longer be replied to.