A Chess Tactics Training Plan by Rating: Volume, Review, and Logging

Twenty minutes a day of properly solved puzzles beats an hour of clicking, and the difference is not effort, it is a rule: you do not touch a piece or a mouse until you have decided on a full line and are prepared to be wrong. Everything below is built on that. Under 1400 the volume is higher and the positions are simpler, because the errors that decide your games are hanging pieces and two-move shots. Above 1400 the volume drops and the positions get long, because by then the errors are calculation errors. In every band you keep a written list of what you missed and you re-solve the misses. That log is the part almost nobody does and the part that makes the rest work.

What counts as solving

A solved puzzle is one where you named the first move, the opponent's best reply, and your follow-up, and then checked. A puzzle where you tried a move to see whether the interface accepted it is not solved. It is a slot machine pull with a chess diagram on it.

This distinction matters because online puzzle interfaces quietly reward the wrong behaviour. The board accepts a move instantly, tells you if it was right, and offers the next one. Nothing in that loop forces you to calculate to the end, and a player can log five thousand puzzles a year having never once held a four-move line in their head.

Two mechanics fix it. On a screen, say the whole line out loud before you move a piece. On paper, write it in the margin. Both feel slow and pedantic. That is the point: the skill being trained is holding a line, and a method that lets you skip the holding trains nothing.

The plan by band

Under 1000. Fifteen puzzles a day, all of them mate in one, mate in two, or win a hanging piece, with no timer anywhere near you. Do not filter by fancy themes yet. The single job at this stage is seeing every capture and check on the board before you decide anything, and the fastest route to it is repetition on positions simple enough that you always find the answer eventually.

1000 to 1400. Fifteen to twenty puzzles, twenty to twenty-five minutes, one session. This is the band where a printed workbook earns its price over an app, because a page cannot be brute-forced by clicking and a covered answer cannot be peeked at accidentally. A banded set such as the 200-puzzle collection pitched at 600 to 1200 is the right shape: fixed difficulty, finite, and finishable, which matters more than most people expect. Solve them in order, on a board, answer covered, and add themed online sets for whatever motif your own losses keep featuring.

1400 to 1800. Fewer, harder, longer. Six to ten positions in a thirty-minute session, and you are allowed to spend eight or ten minutes on one. Write the main line and the critical sideline down before you check. At this band you should also start mixing in positional problems, where the answer is a plan rather than a combination, because your losses stop being one-movers; Silman's Reassess Your Chess Workbook is the standard source for that kind of exercise and forces you to apply an imbalance checklist instead of nodding at one.

1800 and up. Three to five positions, fully analysed, written out in full, then checked against the solution and an engine. Volume has stopped being the variable. What you are training is the discipline of a complete calculation, including the move you did not want to look at, and that only happens on hard positions with a pen in your hand.

Timed modes: what they are for

Puzzle sprint modes are pattern-recognition drills, not calculation drills, and they are good at that narrow job. Three minutes of fast, easy positions genuinely sharpens the instant recognition of forks, pins, and back-rank motifs, which is why players enjoy them and why they are not useless.

They become a problem when they are the whole diet. A player whose entire tactical training is timed sprints gets fast at things they already know and never builds the ability to sit with an unclear position for ten minutes, which is exactly what a long over-the-board game asks for. Keep sprints to one short session a week, and never on a day you are playing a rated game.

Spaced repetition, honestly

The general finding that testing yourself beats re-reading, and that spacing reviews out beats massing them, is about as well established as anything in learning research. What is much less established is how well solved chess puzzles transfer to finding the same idea across the board from a human, under a clock, in a position you have never seen. Nobody should sell you a schedule as if that transfer were measured. Treat the spacing as a sensible way to organise review, not as a guarantee.

The practical version is cheap. Every puzzle you get wrong goes on a list. Re-solve that list the next day, again a week later, and again a month later. Anything you miss three times is not a puzzle you failed; it is a pattern you do not own, and it belongs in a small permanent set you revisit every few months. Most players find that set is smaller than they feared and stubbornly consistent in theme, which is itself the diagnosis.

The error log

Keep it on paper, in the same place you keep your games. A scorebook with room in the margins works, and a larger notation notebook works better if you want your games, your missed puzzles, and your review schedule in one object you can carry to the club.

The entry format that works is three lines and takes twenty seconds:

1. The position reference, book page or puzzle number.
2. What you played and why you liked it.
3. What the answer was, described as a motif rather than a move. "Deflection of the defending rook," not "Rxe6."

The third line is the one that pays. A list of moves is not reviewable. A list of motifs, read back after a month, tells you plainly that you are missing deflections and back-rank ideas and have never once failed to spot a fork, and that is a training plan writing itself.

Where the positions should come from

Three sources, in descending order of value, and most players use only the second.

The best positions are the ones you missed in your own games. A tactic you failed to see against a real opponent, under a clock, in a position you had spent an hour building, carries information no curated set does: it tells you what your blind spot looks like when you are tired and invested. Run an engine over your losses, find the moment it says you had something, and put that position on the misses list before you look at the answer.

The second source is a curated set at a stated difficulty, which is what workbooks and rated online puzzles are. Its advantage is calibration: someone has already sorted the positions so you are not solving a study-level problem on a Tuesday morning.

The third is a themed filter, useful only once you know the theme. Drilling fifty pins is excellent training if pins are your weakness and a waste of a week if they are not, which is why the log comes before the filter.

One warning about positions taken from your own games. Knowing a tactic exists is most of the work, and an engine has already told you that. To get real value, note the position, leave it for a week, then solve it cold with no memory of the verdict.

When to move up a difficulty tier

Use accuracy, not boredom. When you are getting roughly eighty percent of a set right on first attempt without straining, the set has stopped teaching and it is time for harder positions. Below about fifty percent you are guessing, the feedback loop has broken, and you should drop back a tier regardless of what your rating says you should be able to handle.

For online puzzle ratings, resist the urge to protect the number. A puzzle rating that only goes up is a sign you are solving inside your comfort zone. The useful setting is the one where you fail often enough to learn something and not so often that you stop.

How the plan fails

• Streak protection. Once a daily streak exists, the goal quietly changes from solving well to not breaking it, and the last few puzzles of every session get rushed. If you notice this, break the streak deliberately.
• Substituting speed modes for work. Covered above. It is the most common failure and the hardest to see, because the session felt intense.
• Never re-solving. Solving a position once and never returning to it is the same as reading a book chapter once. The misses list is the highest-value twenty minutes in the whole plan.
• Training tactics to avoid studying endgames. Puzzles are the most enjoyable form of chess study, which is why stuck players do them exclusively. If your losses are technical rook endings, more tactics will not touch them.

A week that works

Five days of the main band session. One day of the misses list. One day off, or one day of a short sprint block if you enjoy them. Before a rated game, nothing hard, and nothing timed. After a rated game, the misses list gets whatever you failed to see at the board, which closes the loop between the training and the reason you are doing it.