Training with Puzzles: Why Your Own Mistakes Make the Best Exercises

Puzzle sets are the most popular way to train chess, and most players use them badly: they click the first move that looks good, get told they are wrong, and move on to the next one. Here is how to turn puzzle time into improvement, and why the best puzzles are the ones your own games have already written for you.

By the RookSac teamSeptember 29, 20265 min read

Generic puzzles and personal puzzles

A general puzzle collection trains a general skill: recognising tactical patterns quickly. That is useful, but it has a limit. The puzzle you are served is picked to be interesting to everyone, so you spend much of your time on patterns you already see instantly and only occasionally meet one you actually miss.

A puzzle made from your own game is different in three ways:

• It is a mistake you really made. Every position on the list is one where you had a better move and did not find it. There is no wasted time on things you already know.
• The structure is familiar. It comes from openings you play and positions you reach, so what you learn transfers to your next game far more directly.
• You remember the game. You already know the feeling of that moment, what you were thinking and why the wrong move looked good. That memory helps the correction stick.

This is the same idea as a proper game review, taken one step further: after finding where you went wrong, you return to the position and practise finding the right move until it is automatic.

What makes a good puzzle from a game

Not every mistake makes a good exercise. A position is worth training on when:

• You were not already losing. If the game was lost before the mistake, the exercise teaches nothing about that decision. RookSac only builds puzzles from moments where you were in an equal or better position.
• There is a real difference between moves. If five moves are about equally good, there is nothing to find. A puzzle should have one answer, or a small number of clearly equal ones. RookSac accepts alternatives that reach the same outcome, and asks you to find each of the equal branches rather than guess the engine's favourite.
• It can be solved by thinking. If the answer needs fifteen moves of engine calculation, the puzzle is not teaching a human skill. Skip those.

How to solve a puzzle so that it teaches you

Speed feels productive and mostly is not. The point of a puzzle is not to get it right; it is to practise the process that finds the right move in a real game. That process has five steps:

1. Read the position. Whose move it is, the material, where the kings are, what the last move did.
2. List forcing moves. Checks, captures and threats for you, then for your opponent. Tactics live almost entirely in forcing lines.
3. Calculate to the end. Follow each candidate until something is resolved: a mate, a won piece, a position that is clearly quiet. Then check your opponent's best reply, not their most natural one.
4. Commit to one line before you touch a piece. If you can't state the whole line, you are guessing.
5. Check for a hole. Before you move, ask what your move leaves undefended and what the opponent's most annoying reply is.

These steps are the same ones that stop blunders in real games (see how to stop blundering), which is why puzzle practice done this way helps your over-the-board play and puzzle-clicking mostly does not.

Spend the right amount of time

• Under thirty seconds and correct: the puzzle was too easy for you. Do not count it as training.
• One to five minutes of honest calculation: this is the sweet spot, where you are working at the edge of your ability.
• Over five minutes stuck: use a hint. A good hint points at an idea (a piece, a square, a motif) without giving the move. RookSac's hints are staged for this reason: each one narrows the idea a little further, and none of them hands you the answer.

Getting one right with a hint is worth more than looking up the answer and much more than skipping it. After a hint, try to understand why the idea works.

What to do when you are wrong

This is where most of the learning is, and where most players skip ahead. When your move fails:

1. Find the refutation. Look at the reply that punishes your move. Understand why you missed it.
2. Find the reason the right move works. Not the move: the idea. "The bishop was pinned, so the pawn could not be taken" is an idea. "Qe7+" is only a move.
3. Note the pattern. Fork, pin, back rank, overloaded defender, loose piece. After a few weeks your notes show which patterns you keep missing, and those are the ones to study (see six tactical patterns).
4. Come back to it. Solve the same puzzle again after a day, then a few days, then a week. If you solve it each time without hesitation, you have learned it.

How much and how often

Twenty focused minutes a day beats two hours once a week. The mind consolidates what it has practised repeatedly, and spacing repetitions is one of the best-supported ideas in learning research. A workable routine:

• Five to ten personal puzzles from your recent games, done carefully.
• A couple of repeats of puzzles you missed earlier in the week.
• One short game review, once you have played some new games.

The purpose of the routine is the loop: play, review, extract the mistakes, drill them, then play again and see whether the same mistakes still appear. If they do not, you are improving in the way that matters.

When puzzles are not the answer

Puzzles fix tactical oversight. They do not fix time trouble, a shaky opening repertoire or poor endgame technique, and a report that says most of your losses come from one of those needs a different remedy. That is why the first step is always to find out what your mistakes are before training on them; see reviewing many games at once.