Old Indian Defense: Nf3 — Black's King's Indian Without the Commitment

By playing g6 on move three, you are signalling that your dark-squared bishop belongs on g7, where it will eye the centre and the long diagonal. You haven't yet committed to a kingside fianchetto in a King's Indian sense (no e4 push yet from White), and you haven't blocked in your light-squared bishop. This flexibility is the whole point. If White plays Nc3 (the engine's top choice), you continue with Bg7 and then castle. You'll end up in a King's Indian-style structure — but one where White has already played Nf3, which means the Mar del Plata-style kingside attacks are less dangerous. Your job in the opening is straightforward: develop naturally, castle, and keep an eye on the e5 break. White holds a slight theoretical plus, but in practice you get a very playable middlegame with clear plans.

Stockfish rates this position +0.62, which is a clear edge for White. That means you are slightly worse according to the computer. The engine's preferred move is Nc3, and after that the main line would continue Bg7 e4 O-O. This is the most common path, played over 409,000 times in the database. White scores 48.8% in that line — almost exactly even. So while the computer thinks White has a small plus, the human results show equal chances. That is the Old Indian promise: you accept a tiny theoretical inferiority in exchange for a rich, imbalanced game where your opponent must know what they are doing.

White has several reasonable moves here, but one is a clear mistake. Here are the most-played options and how they treat Black: - Nc3 (409,121 games): The main line. Develop naturally, go Bg7, castle, and aim for e5 or c5 in the middlegame. - g3 (84,198 games): White fianchettos too, leading to a quieter game. Black scores well here — White's win rate drops to 50.3% while Black wins 46.6% — so don't be afraid. - e3 (82,035 games): A solid but passive choice for White. White wins only 44.7% of the time here — that is a terrible score for the first player. Black equalises easily. - Bg5 (74,074 games): This is the mistake. The FACTS show that Bg5 is an inaccuracy — it loses about 0.8 pawns compared to the best move (Nc3). White still wins 47.0% of games here, but you should be the one pressing. The bishop on g5 pins the knight but doesn't accomplish much in this structure; you can simply play Bg7 and castle, or even h6 to ask the bishop a question. - Bf4 (25,264 games): Similar to Bg5 but slightly less committal. Still, White's win rate is just 46.0%.

The numbers from 717,273 games tell a surprising story. White's overall win rate is 48.0%, Black's is 47.7%, and draws are rare at 4.3%. That near-level split is within the margin of error — this position is essentially a 50-50 toss-up in real play, despite the computer preferring White. For comparison, most mainstream openings give White a noticeably higher practical score. The Old Indian with Nf3 is an under-explored, under-theorised line where your opponent is less likely to know the subtleties. If you enjoy positions where you can outplay your opponent in the middle game rather than memorising theory until move 20, this is a great fit.