Sharpen weak spots with goal-driven tasks
A useful tactical task tests one skill rather than accidental agreement with a single engine number. This guide explains how to build sets for mate, preserving an advantage and candidate moves, including positions where Stockfish changes its notation from M12 to a large numeric score.
Define the skill first
“Tactics training” is too broad. One set should test one habit: recognising a mating pattern, calculating a forcing sequence, finding the only defensive move, or simplifying a winning position safely. If every position requires a different skill, the final score does not reveal what improved.
Write the pass condition before creating the set. For mate, preserve the forced attack. For an advantage, stay above the selected threshold. For engine moves, find a line that remains close to the best result after the opponent’s strongest response.
6k1/5ppp/8/8/8/8/5PPP/3R2K1 w - - 0 1Why M12 can become +30
M12 means that the engine currently confirms a forced mate in its principal line. After a move, a depth change or another winning continuation, the display may return to a numeric evaluation. That does not automatically mean the win was lost. A +30 position is practically decisive even if the exact mate length is not confirmed at this moment.
A “preserve the mating net” task should therefore accept two equivalent signals: confirmed mate or an advantage above a safe threshold such as +10 for the attacking side. It must not accept an arbitrary third MultiPV move only because the engine listed it. The move still has to preserve the required result after the best defence.
Control rule
Evaluate the position after the opponent’s strongest response, not only immediately after your move. This rejects attractive checks that let the king escape or lose the advantage one move later.
r1bq1rk1/ppp2ppp/2np1n2/8/2B1P3/2N2N2/PPP2PPP/R1BQ1RK1 w - - 0 8Pattern error
You missed a motif, weak square or piece alignment. Use several related positions and name the motif after each solution.
Calculation error
You saw the idea but stopped before the best defence. Practise shorter lines in which the opponent’s reply must always be stated.
Evaluation error
You calculated the line but misjudged the final position. Compare activity, king safety and material after exchanges.
Pace error
You played the first plausible move too quickly. Require at least two named candidates before moving.
Which error most often ruins your tactical tasks?
- I miss the pattern
- I stop calculating too early
- I misjudge the final position
- I move too quickly
Build a measurable set
Begin with five to ten positions, one finish condition and limited hints. Classify the reason for every miss rather than recording only success or failure. Two accurate runs in a row are a better signal to add depth or material than a single lucky score.
A position is learned only when you can choose the move, predict the best reply and explain why the final position meets the task condition. That process transfers to real games more reliably than memorising one spectacular solution.