Mukhammadzokhid Suyarov vs Artem Uskov: game review

25…Nd5! The only good move here: every alternative is clearly worse.

• The next best move, 25…Rhe8, would drop Black's winning chances from 48% to 32% (+2.03).
• 25…Nd5 threatens …Nxe3, winning the exchange, and stops Bxe6.
• It threatens …Nxe3, winning the exchange, so White has to spend the next move on it.
• White was threatening Bxe6, winning a pawn. Nd5 takes care of it.
• The engine's plan continues with …Bc6 and …Qd8: this move is the first step.
• White answers 26.Bxd5, taking the knight on d5, leaving Black with an isolated pawn on d5.

30…Qc7 is a blunder: Black's winning chances go from 39% to 17%.

• White answers 31.Qc2: adds a decisive attacker on the g6 pawn: 2 attackers against 1 defender.
• After Qc2, White wins 1 point of material within 4 moves.
• Better was 30…Be8 (+1.24): shores up the g6 pawn, which was under as much attack as it had defence (1 against 1).

Better: 30…Be8 shores up the g6 pawn. After 31.Qc2 Bf7 32.Re7 Bxe7 33.Rxe7 Kf8 34.Rxf7+ Kxf7 35.Qxg6+ Ke6, Black comes out a piece up (+1.24).

31.Qc2! The only good move here: every alternative is clearly worse.

• Adds a decisive attacker on the g6 pawn: 2 attackers against 1 defender.
• The queen aims at the g6 pawn, a weak point in front of the Black king.
• The next best move, 31.Qd1, would drop White's winning chances from 83% to 52% (+0.22).
• 31.Qc2 wins a pawn, picking it up with 32.Nxg6, and threatens Qxg6+, winning a pawn.
• Sets up a sequence and wins a pawn once the captures settle, picking it up with 32.Nxg6.
• It threatens Qxg6+, winning a pawn, so Black has to spend the next move on it.

32.Nxg6!! A real sacrifice: White gives up material and the engine still rates it best.

• Puts the knight on g6 where it can be taken.
• Taking is too dangerous: Black's best reply is 32…Re8 33.Nh4 Bf8 34.Nf5+ instead (the engine rates it +4.28).
• Takes the pawn on g6.
• Leaves Black with an isolated pawn on f6: a long-term weakness.
• 32.Nxg6 wins a pawn and threatens Re7, winning the queen for a rook, picking it up with 33.Rxf7+.
• Takes the g6 pawn and wins a pawn once the captures settle.

• The idea: 32.Nxg6!! offers the knight on g6. Taking it gives White the material back and more.
• What White gives up: 32.Nxg6 puts the knight on g6 where it can be taken by the queen on f7. That's a knight on offer.
• If Black takes the knight: 32…Qxg6 is met by 33.Re7+. Then 34.Rxe7+ gives check, and 35.Rxf7+ gives check. Material is level and White is winning (+5.79).
• Black's best try: Black's best is to leave the knight alone and play 32…Re8. White answers 33.Nh4, and the knight on g6 gets away, threatening Nf5+, winning a bishop. Then 34.Nf5+ gives check, and 37.Rxe3 takes the rook on e3. White ends up 6 points of material ahead: White is winning (+4.28).
• Without the sacrifice: The best move without the sacrifice, 32.g3, is clearly weaker (+0.14 against +4.28). Black answers 32…Re8, taking the open e-file, aiming at the rook on e3. That would leave White level.

33.Nh4! The only good move here: every alternative is clearly worse.

• The next best move, 33.Ne7, would drop White's winning chances from 81% to 17% (−4.39).
• 33.Nh4 gets the knight out of the way of …Qxg6 and threatens Nf5+, winning a bishop, picking it up with 34.Nxd6.
• Black was threatening …Qxg6, winning a knight. The knight steps away.
• It threatens Nf5+, winning a bishop, picking it up with 34.Nxd6, so Black has to spend the next move on it.
• The engine's plan continues with Qd1 and Nf5+: this move is the first step.
• Black answers 33…Bf8, winning a rook.

• Both sides followed opening theory up to 9…Ng7: 18 moves in all, 9 by White and 9 by Black.
• Theory runs out after 9…Ng7: too few games between strong players reached this position to go further.