The Italian Game: Nh6 — How to Punish Black's Odd Defence

Castling kingside might look modest, but it is the most principled and effective move here. By safeguarding your king and activating your rook, you keep all your attacking options open. Black's knight on h6 is awkwardly placed — it does nothing to control the centre and can be chased away later with a timely g4 or a pawn thrust. The statistics back this up: from this exact position White scores 60.4% wins with only 3.4% draws. That is a punishing score for Black, and it tells you the Nh6 experiment is not working for them.

Stockfish's best continuation after 4.O-O is d6, followed by d4 exd4 Nxd4. Black tries to shore up the centre with d6, but White immediately strikes back with d4, opening lines. The knight recapture on d4 gives you a lovely outpost. Even against Black's best defence, the engine gives you +1.11 — a healthy edge you can press for the rest of the game. You are not chasing a cheap trick; you are playing solid chess that leaves Black with a lasting disadvantage.

Black has tried several moves here. Here is how you fare against each one, based on real games: Bc5 (26,328 games) — White scores 59.1% against this natural developing move. d6 (13,159 games) — the engine's pick, but you still score 59.6%. f6 (4,781 games) — White does even better at 65.5%. Be7 (4,254 games) — 58.6%. g6 (3,577 games) — 58.3%. a6 (3,193 games) — 64.6%. In every single case you win well over half the games — a reminder that Black is already fighting for equality.

The engine identifies three inaccuracies Black often plays here. Spot them and you gain even more: f6 loses about 0.8 pawns — the correct move was d6. This move weakens the kingside and takes a square from the knight. g6 loses around 0.5 pawns (d6 was better) and creates a hole on f6. a6 costs about 0.6 pawns (again, d6 was right). These are real, concrete errors. If your opponent plays any of them, you can push for a near-decisive advantage with natural, central play — starting with d4 yourself.