How to calculate in chess: a step-by-step method

Calculating in chess means seeing in your head what happens after a move: choose two or three candidate moves, starting with forcing ones —checks, captures and threats—, follow each with your opponent's best replies until the position settles, and judge where you have ended up. It is not about seeing many moves but about seeing the right ones, in order, without forgetting your opponent's.

What calculation is, and what it is not

Almost every move in a game mixes two kinds of thinking. One is general: what the structure is, which pieces are badly placed, what the plan should be. We cover it in how to think in the middlegame. The other is calculation: a concrete sequence in which your opponent's replies are more or less forced and the outcome can be checked.

Calculating is not the same as spotting a tactic. Recognising a pattern —a fork, a pin, the tactical motifs— tells you where to look. Calculating is checking that the pattern actually works here: that your opponent has no in-between check, no piece to interpose, no capture that changes everything. Most calculation errors come not from a lack of imagination but from a lack of checking.

Nor does calculating mean looking at everything. An engine examines millions of positions per second; a human, a handful per minute. Human calculation is therefore, above all, a problem of selection: which moves deserve a look and which can be dismissed straight away.

Step 1: before calculating, ask what has changed

Your opponent's last move is the first piece of information you have. Before thinking about your own ideas, ask:

• What does it threaten? What the piece that moved now attacks, and what it has uncovered behind it.
• What has it stopped defending? A piece that moves abandons squares and pieces it used to guard.
• Which pieces are loose? Yours and theirs. An undefended piece is the raw material of almost every combination.

This takes ten seconds and prevents half of all defeats. Many games that are thrown away from a winning position, as we explain in why do I lose won games, are lost right here: your own plan is calculated beautifully and the other side's threat is never looked at.

Step 2: candidate moves, forcing moves first

Alexander Kotov set out the idea of candidate moves in his 1971 book Think Like a Grandmaster: before you calculate, list the moves that deserve analysis. Two or three are almost always enough; more than five usually means you have not understood the position.

The order in which to look for them is simple and worth turning into a habit:

• Checks. All of them, even the ones that look absurd. A check reduces your opponent's replies to very few, which is what makes a line calculable.
• Captures. All of them, even those that give up material.
• Threats. Moves that attack something and demand an answer.
• Quiet moves. Only then: moves that improve a piece, prevent something or prepare a plan.

Forcing moves come first not because they are always best, but because they are the quickest to calculate and the quickest to change the result. If a capture wins a piece, there is no need to keep searching for the ideal manoeuvre.

The same filter applies to your opponent: when you calculate your line, their candidate replies also start with checks, captures and threats. This line from the Queen's Gambit Accepted shows it:

Black tried to hang on to the extra pawn with 3...b5 and 4...c6. Recapturing on b5 opened the long diagonal, and 6.Qf3 hits the rook on a8, which has no comfortable defence: if 6...Nc6 blocks, the knight is pinned and White wins material. Whoever played 5...cxb5 calculated their own capture, but not the queen move that came next.

Step 3: the tree, one branch at a time

With your candidates chosen, each one is followed by the best reply, and each reply by yours. This is what Kotov called the tree of analysis: a trunk (the current position), a few big branches (your candidates) and smaller branches (the replies).

Three rules help you avoid getting lost:

• One branch at a time. Finish calculating one candidate before moving on to the next. Jumping back and forth is the surest way to waste time and mix up positions.
• Your opponent's best replies, not the ones you hope for. If a line only works when your opponent goes wrong, it is a hope, not a calculation.
• Depth where moves are forced, breadth where they are not. A sequence of checks can be followed for ten moves; a position with many reasonable replies, for two or three.

Plenty of strong players do not follow the method to the letter: sometimes they return to a branch, and sometimes an idea found in the second candidate improves the first. Kotov has been criticised for this, fairly. But as a discipline for anyone starting to calculate seriously, the tree organises your thinking better than anything else.

Step 4: where to stop and how to evaluate

You do not calculate a line to the end of the game. You calculate it until the position goes quiet: when neither side has checks, captures or immediate threats left. At that point you stop and evaluate with general questions:

• Material. Count it using chess piece values, remembering that an active piece is worth more than a buried one.
• King. Whose king is more exposed?
• Activity. Which pieces are playing and which are not?
• Weaknesses. Isolated pawns, undefended squares, an open file towards the king.

Stopping too early is the most common mistake at club level: you see the capture of a piece and stop looking just before the check that wins it back. Stopping too late exists too: calculating fifteen moves of a line that is not forced is wasted effort, because your opponent will deviate long before.

Step 5: the final check before you move

You have chosen your move. Before letting go of the piece, look at the position as it will be and ask the same question as in step one: what checks, captures and threats does your opponent have now? This five-second blunder check saves more games than anything else.

The most famous miniature in the Caro-Kann Defence is a natural move played without that check:

5...Ngf6 develops a piece and looks like textbook play, but it leaves the king with no squares: the knight jumps to d6 with check and at the same time uncovers the queen's check down the e-file. It is double check, and it is mate. One look at the opponent's checks before moving would have avoided it.

Classic calculation mistakes

• Forgetting your opponent's forcing replies. Calculating your sequence as if the other side has to answer the way you expect. The Englund Gambit punishes exactly that:

White played one «logical» move after another —defend, attack the queen, develop— without asking which checks Black still had. The back rank was left unguarded and the queen came in with mate.

• The in-between move. In the middle of a series of exchanges, your opponent does not recapture: they give a check or make a bigger threat, and the count changes.
• Ghost pieces. Calculating with a piece that has already moved or been exchanged in the line. It is a visualisation problem, and it can be trained.
• Misjudging the end of the line. Calculating five moves correctly and then believing a position is good because you are a pawn up, when your king is stripped bare.
• The Kotov syndrome. Spending half an hour on a complicated position, finding nothing clear and, with the clock ticking, playing a move you never examined. The way out is to settle earlier on a reasonable move when the calculation gives no clear answer.
• Calculating what does not need calculating. If the position has no tension, there are no lines to calculate: there is a plan to choose.

How to train your calculation

• Exercises without moving the pieces. Look at the diagram, calculate the full solution —including your opponent's replies— and only then check it. That is the difference between guessing and calculating; in do chess puzzles actually help? we explain how to get the most out of them.
• Write your lines down. In a notebook or a file, with their branches. Seeing the tree on paper helps you organise it in your head.
• Games with time to think. In three-minute games you barely calculate; in games of fifteen minutes or more, you do.
• Check where you went wrong. When you analyse your own games, write down what you calculated at the critical moments before you switch on the engine. Then compare: the mistake is almost always a reply you did not see.
• Study how masters calculate. Watching a grandmaster explain which moves they considered and why they rejected the rest teaches more than the final line. That is what the intuition and calculation classes in Think Better do, and what Alexei Shirov shows in Creativity and Calculation with his celebrated endgame against Veselin Topalov.

If a line from one of your games still does not make sense, Ask the GM lets you ask about that very move.