The result is a clue, not a grade.

Create distance before explanation

Immediately after a match, write one sentence about how it felt and then pause. Frustration, relief, surprise, and confidence can guide attention, but they also shape memory. Naming the feeling makes it visible. A short break gives you a better chance to reconstruct decisions instead of building a case for the result.

Choose one review purpose. You might examine an opening hypothesis, a turn with several lines, a resource exchange, or a role change. Do not try to explain the entire match. A total review often becomes a list of obvious errors, while a bounded review can compare reasons with enough care to produce a reusable lesson.

If you have a replay, hide later turns. If you have only notes or memory, mark gaps explicitly. Never fill a gap with the detail that makes your chosen explanation work. A decision with insufficient evidence should be labelled unresolved and replaced by another review candidate.

Reconstruct the information timeline

For the selected decision, make two columns. The first contains information available before the choice: visible position, prior reveals, known resources, credible opponent range, and your stated plan. The second contains information revealed later. Keep each fact on the correct side even if the later information now feels obvious.

Add assumptions as a separate layer. An assumption is not a flaw by definition. Decisions under uncertainty require them. The review question is whether the assumption was supported, weighted sensibly, and paired with a fallback. An unsupported certainty deserves different correction from a reasonable read that happened to be wrong.

Mark the first moment when new evidence should have changed your plan. This may occur after the reviewed decision. If you adapted then, the original process may have been sound. If you ignored the update, the deeper lesson concerns revision rather than prediction.

• Known then: directly visible or previously confirmed information.
• Assumed then: your weighted model of hidden information and intent.
• Learned later: facts unavailable when the choice was made.
• Update point: the first evidence that should alter the plan.
• Unresolved: a gap too large for a defensible judgment.

Score process before outcome

Judge three layers. Execution asks whether you carried out the intended sequence correctly. Line quality asks whether the chosen option performed acceptably against the range you held. Decision process asks whether you identified the right constraint, considered a real alternative, and committed for a reason tied to the position.

Use plain labels: sound, needs revision, or unresolved. Sound means the reason and line were defensible under the information available, even if the result was poor. Needs revision means a better-supported assumption or line was available. Unresolved means the record cannot support a fair judgment. Avoid numeric scores that suggest a precision the evidence does not have.

Give the best alternative equal detail. Explain its sequence, cost, failure mode, and next position. If the alternative exists only as a cleaner answer after the opponent's response is known, it is not evidence that you missed it then. The comparison must survive when the later cards are hidden again.

Retain one conditional lesson

A review is incomplete until it changes a future behavior. Write the lesson as a condition and a comparison. When the current line spends my only flexible answer for a small gain, I will compare the next credible threat and identify a replacement before committing. This is more durable than be greedier or never waste removal.

Choose one test window, usually the next three to five matches. Record whether the condition appeared, whether you performed the new behavior, and whether it improved the decision. Do not require the behavior to produce wins. Its first job is to expose better information or a more honest comparison.

Archive the rest of the observations without turning them into simultaneous goals. A repeated pattern can become the next lesson after the current test. This keeps review from producing a burden of rules and gives each change enough attention to become natural.