Catalan Opening: a practical guide to the Open and Closed Catalan, with the cost of each move measured

Short answer

The Catalan Opening (1.d4 Nf6 2.c4 e6 3.g3 d5 4.Bg2) is a Queen's Gambit in which the light-squared bishop goes to g2 and presses along the h1–a8 diagonal. This guide measures four Black replies on move four with Stockfish 19, shows how White regains the pawn in the Open Catalan, where holding it with ...b5 drops the a8 rook, which plan keeps the edge in the Closed Catalan, and what changes after the check 4...Bb4+.

The essentials in a few lines

• The Catalan is 1.d4 Nf6 2.c4 e6 3.g3 d5 4.Bg2: the bishop on g2 aims down the h1–a8 diagonal, and 3.g3 instead of 3.Nc3 sidesteps the Nimzo-Indian.
• None of the four measured replies on move four equalises: with Stockfish 19 at depth 22, 4...Be7 leaves +0.19, 4...Bb4+ +0.20, 4...c6 +0.27 and 4...dxc4 +0.31.
• In the Open Catalan, holding the pawn with 5...b5 costs about three tenths (+0.48 against +0.17 after 5.Nf3). In the line 6.a4 c6 7.axb5 cxb5 8.Ne5, 8...Qb6? loses the a8 rook to 9.Bxa8, and the engine measures +7.04.
• In the Closed Catalan, after 7.Qc2 c6, the plan with 8.Nbd2 and e4 measures +0.32; the natural developing move 8.Bf4 leaves only +0.05.
• Differences under a tenth are measurement noise, not theory: on move ten of the Open Catalan, 10.Bf4, 10.Bd2 and 10.Bg5 measure between 0.00 and +0.12, and the choice is a matter of style.

What the bishop on g2 buys

The Catalan is usually described as the opening for players who want a middlegame without shocks. That gets the temperament right and the content wrong: 3.g3 invites Black to take on c4, and the whole position turns on one question, which is who controls the h1–a8 diagonal once the centre opens. The bishop on g2 attacks nothing on move four. It waits, the way a security camera arrests nobody and still changes how people behave in front of it.

Move order matters for a practical reason. With 3.g3 instead of 3.Nc3, White avoids the Nimzo-Indian (3.Nc3 Bb4), and the price is that the check 4...Bb4+ is still available. The same position can also arise from 1.Nf3 d5 2.g3 Nf6 3.Bg2 e6 4.O-O Be7 5.c4 O-O 6.d4, the order for players who prefer not to show d4 early.

We measured four Black replies on move four with Stockfish 19 at depth 22. All of them leave White between +0.19 and +0.31, a narrow band typical of a sound opening. The number that goes against instinct is the order: 4...dxc4, the reply that wins a pawn, leaves White the most, and 4...Be7, the most modest, leaves the least. Taking the pawn is not a mistake; it means accepting that the pawn will come back with positional interest.

Open Catalan: 4...dxc4 and winning the pawn back

The c4 pawn cannot be kept without a concession, which is why the main move order delays the recapture. 5.Nf3 (+0.17) develops and leaves the return to c4 for when it suits; 5.Qa4+ (+0.09) settles the pawn at once. The eight hundredths between them fall inside the engine's own variation from one measurement to the next, and are no reason to prefer either.

Black's side is more instructive. In the main line, ...a6 and ...b5 chase the queen from c4 with gain of time, and ...Bb7 puts a bishop on the same diagonal as the one on g2, looking the other way. That is the structural answer to the Catalan: the long diagonal is not defended with pawns but by occupying it with a piece.

On move ten White chooses a square for the c1 bishop, and the engine has no strong preference: 10.Bf4 measures +0.12, 10.Bd2 +0.05 and 10.Bg5 0.00. In that band the decision is one of style. Bf4 aims at c7, Bd2 is the most flexible, and Bg5 is ready to swap for the f6 knight that controls d5 and e4. Anyone who memorises only one of these will reach the same structure through the other two doors.

Holding the pawn with ...b5: where the diagonal sends the bill

It is the most human mistake in the Catalan. A pawn has been won, and giving it back feels rude. The trouble with ...b5 is not the pawn but what it leaves behind: with the pawn gone from b7 and the c6 pawn exchanged on b5, nothing stands between the g2 bishop and the a8 rook.

The sequence 6.a4 c6 7.axb5 cxb5 is the whole mechanism. The a4 advance hits the pawn chain at its base, the exchange on b5 removes the c6 pawn from the diagonal, and 8.Ne5 moves White's own knight out of the bishop's way: on f3 it was blocking the diagonal that now runs from g2 to a8 with nothing in between.

With the rook in the line of fire, Black needs a move that closes the diagonal, and 8...Nd5 does it with a piece (+0.60). 8...Qb6, which defends b5 and looks consolidating, ignores the bishop. After 9.Bxa8 the engine puts White more than seven pawns up in evaluation: that is no longer an opening edge, it is a game decided on move nine.

Closed Catalan: the plan with Nbd2 and e4

In the Closed Catalan Black does not take on c4 and holds d5 with ...c6, in a structure that resembles the Slav. With no pawn to regain, White needs another target, and it is the centre: the g2 bishop only sees the whole diagonal again once the d5 pawn is exchanged, and the move that forces the issue is e4.

This is where automatic development is expensive. 8.Bf4 puts the bishop on a good diagonal and leaves the position at +0.05, practically the value of an equalised opening. 8.Nbd2 looks more modest, shuts in the c1 bishop for a while and measures +0.32, because it supports e4 and keeps c4 covered by the knight.

After 8.Nbd2 b6 9.e4 the decision passes to Black, and the most natural move is the most costly: 9...Bb7 measures +0.52, against +0.28 for 9...dxc4, which releases the tension before the centre locks. A similar warning applies one move earlier. With 7...b6 instead of 7...c6, the evaluation rises from +0.33 to +0.68: without a pawn on c6, d5 loses the support this structure relies on.

The check 4...Bb4+

The check wins nothing by itself. It asks which piece White wants to block with, and each answer shifts the position slightly. 5.Bd2 is the block that develops; 5.Nd2 leaves the c1 bishop free and covers c4 and e4 with the knight. Both sit in the same evaluation band, and choosing between them is a matter of taste.

After 5.Bd2 Black has three options, and the retreat is the most accurate: 5...Be7 leaves +0.05, 5...a5 +0.21 and the exchange 5...Bxd2+ +0.30. The retreat looks like a waste of time, since the bishop went to b4 only to come back to e7. The engine prefers it because the exchange helps White: after 5...Bxd2+ 6.Nxd2 the knight reaches d2 without spending a move of its own.

The bigger difference comes two moves later. With White castled, in 5...Be7 6.Nf3 O-O 7.O-O, 7...dxc4 measures +0.37 and 7...c6 +0.12. Taking the pawn in this order hands White the same recovery as in the Open Catalan, with Qc2 and Qxc4, without Black having spent time on ...a6 and ...b5.

Who the Catalan suits, and how to study it

The margins measured here sit almost entirely between +0.1 and +0.3. That means the opening rarely wins the game by itself: it delivers a position in which the side that understands the long diagonal makes fewer mistakes. It is a trade that pleases players who like to manoeuvre and disappoints those who expect to leave the opening with an attack ready.

The efficient way to study it is the one in our guide to building an opening repertoire: fix the three structures, the Open with ...a6 and ...b5, the Closed with ...c6 and the one after the check on b4, and solve the critical positions of each before memorising move orders. When one of your games leaves what you studied, the question to ask is always the same: who controls the h1–a8 diagonal after the next exchange in the centre?

With an engine, one rule from this guide applies to every opening. A difference under a tenth between two moves is noise: in two measurements of the same position, with the same Stockfish at the same depth, the evaluation after 4...Bb4+ 5.Bd2 came out at 0.00 once and +0.12 the other time. What deserves study are the jumps of three tenths, like those of ...b5 and 8.Bf4, and the jumps of seven pawns, like 8...Qb6?.

Frequently asked questions

Openings mentioned in this guide

• Catalan Opening (E00)
• Queen's Gambit Declined (QGD) (D30)
• Nimzo-Indian Defense (E20)
• Bogo-Indian Defense (E11)
• Slav Defense (D10)