vinidivici: I like the move 23.Ndf5.
You need to consider many probabilities. And i dont like to move the interface back and forth looking for fine solutions. Let the brain works. So its like the real match. I only move back and forth the interface to find the solutions if its a complex line and that for save the time.
23.Ndf5 is a tricky move. Looks hard because there are many replies.
Bxb2/gxf5/or even Bxg2.
This is example of analytical calculations. I just share a bit of that, so you probably know how a GM thinks.
The point Ndf5 is to open the d5 file for the rook so pinning the d-6 knight. The risks are the 1.unprotected b2 bishop and 2.giving the f5 knight to the pawn.
Just like u have seen 23...Bxb2 replied by 24.Bxb7
24...Qxb7(to save the rook) and materials back with
25.Nxd6 Nxd6 26.Rxd6
How about 24...Rb8, now the b7 bishop and f5 knight are vulnerable. But its easy to think.
The simples is 25.Bxc8 25...Qxc8/Rbxc8 useless because Ne7+ (forking)
And there is 25.Ne7+, deserves to watch 25...Nxe7 26.Rxe7 threaten the queen and after queen retreat e.g 26...Qd8 then b-7 bishop can retreat so at least its SAFE. The unprotected b2 bishop its NOTHING.
We come to the f5 knight problem.
If 23...gxf5 (lost the knight) then there are 2 choices 24.Bxb7 or Bxg7
24.Bxb7 Qxb7 25.Bxg7 Kxg7 then 26.Nxf5+ after the king retreat and takingd-6 knight, withe up a pawn. So i guessed 24.Bxb7 isnt bad.
24.Bxg7(threaten the rook) if 24...Bxg2 then 25.Bxf8 white wins the exchange bishop traded with rook.
So after Bxg7 24...Kxg6 left. After 25.Nxe5+ 25...Kg8 26.Bxb7 Qxb7 then 27.Nxd6
again, its a good line all materials are back and even a pawn up for white.
So, the conclusion is 23.Ndf5 is a good move.
Maybe there are mistakes on my calculations, u can correct it yourself. And for the GM level these kind of analysis needed even for A PAWN.
