Bc2, Qf3, h4 or f4? How to choose in one London position

ChessInsight team · 15 September 2026

Black's c8 bishop is shut in and the attack is there. A 1,350-chess.com rated student's games show when to play Bc2 or Qf3, and when h4, g4 or f4 are right.

This is the story of one student: a club player rated about 1,350 in 15-minute online games, who plays the London System as White. The student loves the attacking set-up of bishop on d3, knight on e5, then an attack on Black's king. One position from the student's own games kept coming back, and it holds the whole lesson.

The position

Nine moves in, White has everything the attack needs. The bishop on d3 aims at h7, the knight sits on e5, and the other bishop controls the centre from f4.

Black's bishop on c8 is the reason the attack is real. Black played …e6 before moving it, so it is shut in behind its own pawn on e6 and can't fight for the light squares around its king. Nothing on Black's side can challenge White's bishop on d3.

White to play after 9…Bc7. Black's c8 bishop (red) is shut in behind e6, so White's d3 bishop (green) has no rival. What would you play?

Solve this position from the student's game on ChessInsight →

The engine rates this position about +1.3 for White. The student saw an attack and played 9.g4. After Black's best reply the game was level.

The attack was the right idea. The move order was wrong. White's choices here are Bc2, Qf3, a pawn push with h4 or g4, and in some versions f4. Which one is right depends on three questions.

Question 1: Can Black swap on e5, and where is Black's king?

Here Black's knight on d7 is ready to take on e5, and Black's king is still on e8. In that case, bring pieces first.

White's move 9	Engine	What happens
9.Bc2	about +1.3	Keeps the attacking bishop on the b1–h7 diagonal and out of reach of a knight coming to c5 or e4
9.Qf3	about +1.2	The queen joins the kingside first. After …Nxe5 dxe5 Nd7, it goes on to g3
9.O-O or 9.Ndf3	about +1.0 to +1.1	Solid development that keeps the advantage
9.h4	about +0.3	Black swaps on e5, and the pawn has no target yet
9.g4	about +0.1	The same swap, and it loosens White's own king as well

A London System course makes the same point about Bc2 in a similar position: it keeps "our main attacking piece alive".

This was not a one-off. In another of the student's games, with Black's king also still in the centre, the engine's first choices were again Qf3 (+0.96) and Bc2 (+0.92). g4 was never better than level.

After 9.Bc2: what if Black doesn't take on e5?

Black doesn't have to swap. Maia is a human-move model that predicts what players at a given rating tend to play. For players around 1500, it has 9.Bc2 answered by …O-O 29% of the time and by …Nxe5 20% of the time. The rest spread across several moves, and White's plan depends on which one Black chooses.

9…O-O (29% at 1500): Black's king has castled, so the pawns go. Play 10.h4, or 10.Qf3 first. About +1.6.
9…h6 (7%): 10.Bg3, and after …Nxe5 dxe5 Nd7, f4 (see Question 3). About +1.3.
9…c5 (7%): keep building on the kingside with 10.Qf3 or 10.h4. About +1.5.
9…Bxe5 (7%): take back with the pawn, 10.dxe5, so the f6 knight has to retreat, then Qg4. About +2.4, while 10.Bxe5 is only +1.1.
9…Nh5? (7%): 10.Qxh5. The knight is simply loose. About +4.8.
9…Qe7 or 9…b6 (3–4%): look at the knight jumps first, 10.Nxc6 or 10.Nxf7. About +1.6 after …Qe7 and +5.7 after …b6.
9…g6 or 9…h5 (under 3%): 10.Ndf3, backing up the e5 knight with the other knight. Against …h5, don't push h4 or g4. About +1.2 and +1.4.

Three things hold in every line:

Keep the e5 knight. If Black won't take it, support it with the other knight rather than moving it.
Don't open the centre with e4. It scored between +0.1 and +0.9, against +1.2 to +1.7 for the plans above.
g4 is the wrong pawn. It never scored better than +0.5. h4 was a reasonable second choice, except against …h5, and the best move only after Black castled.

The student met exactly this in another game. After 10.Bc2 Nc6 Black declined to swap, and the engine's first choice was 11.Ndf3 (+0.96). The student played 11.g4 instead. After 11…Bd7, the knight on e5 had 12.Nxf7 (about +2.7). Puzzle 3 below comes from the same game.

Question 2: Has Black castled? Then the pawns go

The game went on 9…Nxe5 10.dxe5 Nd7 11.Nf3 O-O 12.Qc2 g6. Once Black's king had gone to g8, the pawn storm became the right plan. At move 13 the engine's first choice was exactly what the student played.

The same game, move 13. Black has castled and the c8 bishop is still shut in. Now 13.h4 is the engine's first choice (about +2.0), and the student played it.

The engine confirms it from the other direction too. Take the move-9 position and let Black castle instead of playing …Nd7: now 9.h4 becomes the best move (+1.29), ahead of Bc2 (+0.93) and Qf3 (+0.97). In a third game of the student's, with Black castled, the engine's first choices were h4 (+1.77) and then h5 (+1.97).

A London course plays h4 in this structure too, as preparation for bringing a knight to g5 and opening the h-file.

h4 or g4? g4 asks for more than h4. It was only strong when Black had castled and had no knight ready to swap on e5: +1.68 in that version of the position. Once White has castled short as well, g4 mostly weakens White's own king. The engine scores it 0.00 there.

Question 3: Has your f4 bishop moved? Then f4 can hold e5

f4 is impossible while White's bishop stands on f4. It becomes a plan when that bishop has dropped back.

Take the same position after …h6, where the engine's main line is 9.Bg3 Nxe5 10.dxe5 Nd7. Now 11.f4 is best (about +1.25). The pawn on e5 is secured and Black's pieces stay cramped. 11.Nf3 is only +0.47, and 11.h4 is about −1.4.

If Black then breaks with …f6, 12.Bg6+ punishes it (about +2.3).

A London course notes that White "can easily support the knight with f4". The same course warns that played at the wrong moment, f4 allows …f6 and Black equalises. So f4 is a follow-up once the centre is settled, not a first move.

The rule to take to the board

With Black's c8 bishop shut in: pieces first, pawns once the king has castled.

Black's king in the centre and a knight ready to take on e5? Bring pieces: Bc2 or Qf3.
Black has castled? Push h4, then h5. Choose g4 only when Black can't swap off your e5 knight and your own king isn't castled short.
Your f4 bishop gone to g3? f4 holds e5, once the centre is settled.

Your turn: three positions from the student's games

All three have Black's light-squared bishop shut in behind e6. Each shows White to play, with the student's move in red. Find something better, then check yourself on the real puzzle.

What a coach would add

The analysis found the position in the student's own games, checked each plan with the engine, and matched it against what London System books recommend. A coach can take it from there: how to practise the three questions until they are automatic, what to do when Black breaks with …c5 or …f6 early, and whether this attacking style suits the student. Those conversations start in a much better place with the student's own games on the table.

AI finds the moments. Coaches bring the judgement.

Try it with your own games: upload a game to ChessInsight and it turns your real mistakes into puzzles you can train with.