Jorden Van Foreest vs Rauf Mamedov: game review

33.Nxd6!! A real sacrifice: White gives up material and the engine still rates it best.

• Leaves the rook on c7 where it can be taken.
• Black can take, but after 33…Kxc7 34.Nxf7 b5 35.Bxd5 the engine rates it +0.45.
• Takes the knight on d6.
• Attacks the f7 pawn, which is undefended.
• 33.Nxd6 trades off the attacked knight instead of losing it to …Rxb5.
• Black was threatening …Rxb5, winning a rook for a pawn. The knight takes something with it instead.

• The idea: 33.Nxd6!! gives up the rook on c7 to stay in the game: after it White holds the balance (+0.45); anything else leaves White a little worse (−1.44).
• What White gives up: 33.Nxd6 leaves the rook on c7 hanging to the king. That's a whole rook on offer.
• What it does instead: It takes the knight on d6. The knight on d6 was the only piece guarding c8 and e8 next to Black's king.
• If Black takes the rook: 33…Kxc7 is met by 34.Nxf7, taking the f7 pawn. Then 35.Bxd5 takes the d5 pawn, and 37.Ne5 attacks the g6 pawn, which is undefended. White is a pawn down: White holds the balance (+0.41). This is Black's best try.
• Without the sacrifice: The best move without the sacrifice, 33.Bxa4, is clearly weaker (−1.44 against +0.45). Black answers 33…Rxa4, taking the bishop on a4 and threatening …Rxa2+, winning a knight. That would leave White a little worse.

42…Kd6 is a blunder: Black's winning chances go from 38% to 17%.

• White answers 43.Nxg6, which takes the pawn on g6.
• After Nxg6, White wins 1 point of material within 2 moves.
• Better was 42…Rc6 (+1.31), which protects the g6 pawn, which was hanging.

Better: 42…Rc6 stops Nxg6. Engine line: 43.Bd5 Ra6 44.Nf7 Nb6 45.Bb3 Nd7 (+1.31).

46.g6! The only good move here: every alternative is clearly worse.

• Pushes the passed pawn to g6, close to promotion.
• The next best move, 46.f5, would drop White's winning chances from 90% to 57% (+0.78).
• 46.g6 pushes the passed pawn to g6.
• Black answers 46…Rc3.
• After 46…Rc3 47.f5 Ke7 48.Nd3 Ra3 49.Nxb4 Ra4, White comes out a pawn up (+4.99).

47.g7! The only good move here: every alternative is clearly worse.

• Pushes the passed pawn to g7, close to promotion.
• The next best move, 47.f5, would drop White's winning chances from 90% to 53% (+0.29).
• 47.g7 pushes the passed pawn to g7.
• Black answers 47…Rg1, attacking the g7 pawn, which is undefended.
• Engine line: 47…Rg1 48.Nf7+ Kc6 49.Ng5 Rd1+ 50.Kc4 (+6.18).

48.Nf7+! The only good move here: every alternative is clearly worse.

• Gives check, so the reply is forced.
• The next best move, 48.Nd3, would drop White's winning chances from 91% to 50% (+0.01).
• 48.Nf7+ wins the exchange, picking it up with 50.fxg5.
• Sets up a sequence and wins the exchange once the captures settle, picking it up with 50.fxg5.
• Black answers 48…Ke6, attacking the knight on f7, which is undefended, leaving White with doubled g-pawns.
• After 48…Ke6 49.Ng5+ Rxg5 50.fxg5 Kf7 51.Kc4 Kxg7 52.Kxb4 Kg6, White comes out two pawns up (+6.09).

• Both sides followed opening theory up to 6…Qxe7: 12 moves in all, 6 by White and 6 by Black.
• Theory runs out after 6…Qxe7: too few games between strong players reached this position to go further.