How to calculate: a thinking method for every move

Published 26 October 2026

Most games below master level are not decided by deep strategy. They are decided by one move that was played without enough checking: a piece left hanging, a mate in one missed, a capture that walked into a trap. Calculation is the skill that prevents those moves. It sounds like a talent, something you either have or do not, but most of it is a method. If you follow the same short routine every time you face a real decision, you will see more and blunder less, even before your "vision" improves.

This article describes one such routine and works through two small positions. The move sequences and the claims about them (which move is legal, which reply is forced, which final position is mate) were checked with a chess rules library.

Step 1: what changed?

Before thinking about your own move, look at your opponent's last one. What does the moved piece attack now? What did it stop defending? What line did it open behind it? A surprising number of losses come from players who are busy with their own plan and never asked why the opponent played that move. This step takes a few seconds and catches most simple threats.

Step 2: list candidate moves

A candidate move is a move you will consider seriously. The idea, popularised by the Soviet trainer Alexander Kotov, is to write a short list of two to four candidates before analysing any of them deeply, rather than following the first idea that comes to mind and then jumping between half-analysed lines. The list stops you from falling in love with one move, and it lets you compare options instead of judging a single move in isolation.

Where do candidates come from? The most reliable source is the next step.

Step 3: checks, captures and threats

Forcing moves limit the opponent's replies, which makes them both stronger and easier to calculate. A good habit is to scan, in order:

1. Checks: every check you can give, even ones that look silly.
2. Captures: every capture, including ones that give up material.
3. Threats: moves that attack something valuable or threaten mate.

Then do the same scan for your opponent: if it were their move, what checks, captures and threats would they have? That second scan tells you what you must not allow.

Here the obvious move is to take the queen. Both queens are attacking each other, and White moves first. But run the scan for Black. If White plays Qxd7, what checks would Black have? Rxe1 is a check, and it captures the white rook, which is defended by nothing. The white king has no escape square because of its own pawns on f2, g2 and h2.

1.Qxd7 Rxe1#

Now scan White's own checks. Rxe8 is a check, and it captures a rook. Black's only legal reply is Qxe8, because the black king cannot move and nothing else can take the rook. That recapture takes the black queen off d7 and onto e8, where the white queen can reach it.

1.Rxe8+ Qxe8 2.Qxe8#

The two candidate moves both win material at first sight. Only the routine of checking the opponent's forcing replies shows that one loses and the other mates. In a real game the difference is rarely this dramatic, but the method is the same.

Step 4: count the exchanges

When pieces can be captured on a square, count attackers and defenders, and then the values of the pieces involved. A piece attacked twice and defended once can usually be won, but only if the cheap attackers capture first.

1.Nxd5 exd5 2.Bxd5

Two attackers against one defender, so White comes out ahead. The order matters little here, because a knight and a bishop are worth about the same. It would matter a great deal if one of the attackers were a queen: capture with the queen first, and the pawn recaptures and wins it. The rule of thumb is to use your least valuable attacker first and to stop counting only when no more captures are possible.

Step 5: the blunder check

Once you have chosen a move, and before you play it, imagine it on the board and ask one more time: what checks, captures and threats would my opponent have after this move? This is the step most often skipped, especially when the move feels obviously right. It is also where the example above would have been saved: Qxd7 looked obviously right.

The blunder check takes ten or twenty seconds. In fast games you cannot do it every move, but you can do it whenever you make a capture, move a piece that was defending something, or move a pawn in front of your king.

How deep should you calculate?

Less deep than you think. Club players often try to see ten moves ahead in a position where three would decide the issue, and then lose a piece to a two-move tactic. Calculate forcing lines until they stop being forcing, that is, until there are no more checks, captures or direct threats, and then judge the position you reach. Quiet positions are better handled by the planning questions in the strategy article than by long calculation.

Two practical tips help. First, calculate with your eyes on the actual board, not on an imagined one, and name the squares to yourself as you go. Second, if you lose track of a line, start it again from the beginning rather than guessing where the pieces are.

A routine to practise

• What did the opponent's last move change?
• What are my candidate moves? Start with checks, captures and threats.
• For each candidate: what are the opponent's forcing replies?
• Count any exchange to the end, cheapest attacker first.
• Before moving: blunder check.

Play a few slower games on this site using the routine on every move that is not obvious. It will feel slow at first. After a few dozen games most of it becomes automatic, and the time it costs goes down while the number of pieces you lose goes down even faster.