Deblunder study notebook
A chess training plan for busy players
A training plan is not a list of everything you should study. It is a small set of decisions about what to stop repeating. If you have thirty minutes a week, the plan has to fit in thirty minutes. This guide shows how to build one from your own games, with a weekly structure, a four-week block and signals that tell you whether it is working.
Why most training plans fail
The usual plan is a pile of activities: tactics, openings, endgames, videos and puzzles. Each activity is useful in isolation, but together they do not answer the question that matters: what mistake am I trying to stop making? Without that target, more study produces more familiarity and little transfer to games.
The second failure is measurement. Rating moves slowly and depends on opponents, so it is a poor weekly signal. A plan needs a signal you can observe in a few games: does the named cause appear less often? Can you solve the positions cold? Did the first-attempt result change?
The third failure is volume. A plan that requires two hours a day survives for a week. A plan that requires twenty focused minutes, three times a week, is more likely to be repeated long enough to produce a sample.
Step 1: audit a real sample
Choose a sample of completed games in one time control. Ten to twenty games is enough to see a pattern without turning the audit into a project. Keep rapid, blitz and bullet separate, because the decision conditions are different. If you play several platforms, note where each game came from.
Review the games in a fixed order. Start with your own explanation of two or three critical moments, then verify with an engine. Group the mistakes by the relation on the board, not by the label. "Lost a rook" is an outcome. "Moved the only defender of a knight" is a cause you can check next time.
Choose the cause that appears in more than one game and that you can describe in one sentence. If two causes tie, choose the one that is easier to observe during a game. A cause you can catch before moving is more useful than a subtle evaluation error you cannot recognize yet.
- One time control, ten to twenty games.
- Two or three critical moments per game, not every move.
- A written cause for each moment, in your own words.
- A count of how often each cause repeats.
- One cause for the next block, not five.
Step 2: choose a measurable target
A target such as "improve my chess" cannot be evaluated. A target such as "in the next ten rapid games, check what my moving piece stops defending before every capture" can. The target should describe an observable behaviour, not a rating number.
Pair the behaviour with a training set. If the cause is moving a defender, collect five positions from your games where that happened. Those positions become the practice material. If the cause is calculation, collect the positions where you stopped the line too early. The set is small, specific and yours.
Step 3: fit the plan to your week
The structure below uses three short sessions. Adapt the content, not the rhythm. The rhythm matters because it alternates play, review and recall, which are different skills. Use the clock honestly: if you have thirty minutes, do not schedule sixty.
Thirty minutes a week: one twelve-minute diagnosis of a single game, one twelve-minute training set of three positions, and six minutes writing the check you will use in the next game. It is a small plan, but it is repeatable.
Sixty to ninety minutes a week: one longer review of two games, two short training sessions of three to five positions, and one recall session with the answers hidden. Keep one day free of chess study.
Three hours or more: add a second review block and a deliberate practice block for a different skill, such as calculation or endgames. Do not add a second cause until the first one has a stable sample.
- Session A: review one completed game and name one cause.
- Session B: solve three to five positions from that cause, answer hidden.
- Session C: recall the same positions after a few days, then play.
- Every week: write one sentence describing the check you will use in games.
- Every four weeks: compare the cause count and the first-attempt results.
A four-week block
Week one is the audit. Collect the games, choose the cause and build a set of five positions. Do not try to fix anything yet. Week two is the first training pass. Solve the positions cold, write your reasoning and compare it with the verified continuation. Week three is transfer. Play a small number of games and look for the moment the check applies. Week four is review. Count how often the cause appeared and decide whether to repeat the block or change the target.
This block is intentionally short. A cause that changes in four weeks gives you a next decision. A cause that persists needs a different representation or a smaller target, not more of the same volume.
What to track
Track the process, not only the result. The exact counts will be small, so report them as counts and comparisons rather than percentages with false precision. A monthly view is more stable than a weekly one.
- How many times the named cause appears in the sample.
- How many positions you solved on the first attempt, without hints.
- Whether you recognized the check during a game, even if the move was not perfect.
- The number of games in the sample and the time control.
- Your own note about what made the cause easier or harder to see.
How to adjust without starting over
If the cause appears less often, keep the check and choose a second cause for the next block. If it appears at the same rate, change the training representation: use positions from the opponent's side, remove the annotations, or shorten the decision time to simulate a game. If it appears more often, check whether you are now noticing it, which can happen before the behaviour changes.
Do not change every variable at once. One cause, one training set, one measurement window. If the plan feels too easy, add a second recall day before adding a new topic. If it feels too hard, reduce the set to three positions and keep the weekly rhythm.
Where Deblunder fits
Deblunder is built around this loop. It imports completed games, analyzes them and looks for causes that repeat across games. You can see the sample behind a priority instead of treating one loss as a trend. Training uses 3, 5 or 10 positions from your games, with Solve, Understand, Compare and Recall, and only the first valid attempt is scored.
The first diagnosis and first training are free without a card. Continued full access has paid conditions in the app. The tool does not replace your plan or guarantee a rating gain. It gives the plan a source of positions and a visible sample. For the full method and limits, read how Deblunder works.
Frequently asked questions
How long should a chess training session be?
Short enough to repeat. Twenty to thirty focused minutes, three times a week, can support a review, a small training set and a recall session. A longer session is useful only if you can sustain it consistently.
Should I study openings, tactics or endgames first?
Start with the cause that costs you the most points in your own sample. If you do not have a sample, review ten to twenty games before choosing. A generic priority list is less useful than a pattern you can see in your games.
How many games do I need before I can see a pattern?
Ten to twenty games in one time control is a reasonable starting sample for a personal review. It is not a statistical guarantee. Report the count, keep the sample visible and treat the pattern as a hypothesis to test.
What if my rating does not change?
Rating depends on opponents, time controls and sample size. Use process signals such as cause recurrence and first-attempt performance, and give the plan several weeks before judging it. A stable rating does not prove the training failed, and a rising rating does not prove it caused the gain.
Can I train without an engine?
Yes for the review and recall structure. An engine helps verify consequences and find candidate mistakes, but you can start with your own notes and a board. The important part is comparing your reasoning with a verified continuation.