How to find tactics in chess

Strong players do not calculate more than you. They calculate less, because they know which moves deserve a look. Most of a position is quiet, and a tactic is a needle in that hay: one forcing move that changes everything. The trick is that forcing moves are a tiny set. In a typical middlegame there are thirty to forty legal moves, and rarely more than five that give check, capture something or make a direct threat.

So the method is mechanical. Before you think about plans, list the forcing moves for both sides. Look at each one for a few seconds, not to calculate it to the end, but to see whether it changes the geometry of the position: a king that loses a defender, a piece that lands on two targets, a line that opens. If nothing happens, play your plan. If something happens, calculate that one move properly.

The list below is ordered by how often each question finds something. Checks first, because a check restricts the reply to a handful of moves and that is where combinations live.

The classic trio, in this order, for both colours.

• Checks. Every check, including the ones that look absurd. A check that costs a piece is the start of most sacrifices; a check that gains a tempo is the start of most forks.
• Captures. Every capture, including the ones that lose material on the spot. Chess is full of captures that look bad for one move and good on the next, because the recapture pulls a piece onto a square where it is hit again.
• Threats. Moves that attack an undefended piece, threaten mate or attack the queen. A threat forces a reply almost as well as a check does.

The first diagram shows why checks come first. White's knight is doing nothing on d5, and there is no obvious target. But there is a check, Nc7+, and the moment you look at it you see that c7 also hits the rook on a8. The king has to move, and the rook is gone. Nobody finds this by scanning for forks. Everybody finds it by scanning for checks.

A loose piece is a piece that nobody defends. The rule of thumb, loose pieces drop off, is the single most reliable tactical signal in amateur chess, because a loose piece turns an ordinary attacking move into a double threat. Attack a defended piece and your opponent shrugs. Attack a loose piece together with anything else, a king, a second loose piece, a mate threat, and something falls.

So on every move, count the loose pieces on both sides. Yours first: if you have two, the opponent may already have a fork available and you should fix that before you dream. Then theirs: two loose pieces on the same colour of square are a knight fork waiting for a knight; two loose pieces on a line are a rook or queen fork waiting for a check that puts your piece on that line.

A pinned piece counts as loose even when it is defended, because it cannot run. The second diagram is the most common version. Black's knight on e5 stands between White's rook and the black king, so it is pinned and it cannot move. It is defended by the d-pawn, but that does not help: White attacks it a second time with a pawn, and the knight has no escape. A pawn for a piece, found by asking one question: what cannot move?

Look at each king and ask which lines lead to it: the files, the ranks and the diagonals that would give check if the pieces in between moved. A king with one open line is a target. A king with a line that is blocked only by one of your own pieces is a target on your next move, because that piece can leave with a threat and the check arrives behind it.

That is the discovered attack, and it is the most violent pattern in the checklist, because the moving piece is free to do anything: capture the queen, fork two pieces, attack a rook. The opponent has to deal with the check first, and the moving piece is collected on the following move.

In the third diagram the black king sits at the end of the b3-g8 diagonal, and the only thing between it and White's bishop is White's own knight. The knight is free to go wherever it hurts most. It takes the queen on c7, the bishop gives check, the king steps aside, and the knight is still there to take the rook. Before you calculate anything in a position like this, notice the diagonal. The tactic is a consequence of the line, not the other way round.

The last question is quieter and finds the tactics that are not on the surface. For every important target, the king, a loose piece, a mating square, ask: who defends it, and what else is that defender doing? A piece that guards two things at once is overloaded. Force it to do one job and it abandons the other.

The tools are the deflection, where you capture or attack the defender so that it has to leave its post, and the removal of the defender, where you simply trade it off. The check-capture-threat scan finds these too, but only if you know what to look for: a capture that appears to lose material is often a deflection, and the material comes back with interest one move later.

A practical way to see overloading: mentally remove each defending piece from the board and ask what would fall. If the answer is "the queen" or "mate", find a forcing way to remove it for real.

Reading the list is worth nothing. Running it a few thousand times is worth several hundred rating points, and there is a specific way to practise it.

Solve puzzles where the motif is not announced. A puzzle titled "fork" trains you to find forks when someone tells you there is one; your games never will. Mixed-motif puzzles at your rating force the scan, and a scan that works when the theme is hidden is the one you will actually use over the board. That is why the gym never shows the motif before the solve, and why every miss in the scan of your own games is tagged after the fact: you saw the position cold, exactly as in the game.

Second, do it in your games, slowly, on every move, for a month. It will feel tedious for a week and then it will disappear into your thinking, which is the point. The check is the first move you look at; a loose piece registers as a smell before you count anything; a king on your bishop's diagonal makes you look at what stands in between. When that happens you are not finding tactics any more. You are seeing them.