Turn a replay into evidence, not a verdict.

Prepare a private, bounded record

Use a replay you are permitted to view. Keep it local and avoid recording player identifiers, account names, chat, or unrelated personal details. Decision Fieldbook needs the game state and your reasoning, not a public profile. If you share a study note later, remove identifiers and any link that exposes another person's account history.

Set a forty-minute limit. Write the match's broad plan in one sentence, then state the skill you intend to examine. A focused purpose might be opening discipline, turn sequencing, resource timing, or plan revision. The time limit and purpose protect the session from becoming an exhaustive search for every imperfect action.

Create three note columns: fact, assumption, and later evidence. Add a fourth line for unresolved gaps. You can use paper or plain text. No tracker or special platform is required, and no replay data should be uploaded merely to complete this method.

Use two light passes before one deep pass

On the first pass, watch the match at normal speed. Mark only moments where you hesitated, acted automatically, changed the plan, spent a unique resource, or felt surprised by the result. Do not solve them yet. The purpose is to locate uncertainty without letting the first interesting turn consume the session.

On the second pass, inspect the marks and choose the earliest decision with enough information to reconstruct. Prefer a moment that influenced later options and contains at least two plausible lines. Reject a candidate if memory or replay visibility leaves a critical gap. A later dramatic turn may be less useful than an early commitment that made the drama unavoidable.

Reserve the remaining time for a single deep pass. Hide later information, copy the position facts, state the decision, and compare the chosen line with the best alternative. If you finish early, improve the alternative rather than adding more review targets.

1. Five minutes: set purpose, privacy boundary, and broad plan.
2. Eight minutes: first pass and uncertainty marks.
3. Five minutes: second pass and candidate selection.
4. Seventeen minutes: reconstruct and compare one decision.
5. Five minutes: write one lesson and its next test.

Run the decision trace on the chosen moment

Observe the position with the replay stopped before the choice. Record visible resources, board, hand, health pressure, known information, opponent range, and the role you believed you occupied. Mark any fact that you did not actually notice during play. That distinction tells you whether the issue was knowledge, attention, or comparison.

Name the decision as a tradeoff. Compare two complete lines with order, cost, resulting position, common favorable response, and credible punishment. Identify the binding constraint and the resource each line preserves or spends. Then restate the original reason as fairly as possible, even if the line now looks weak.

Reveal the following turns in stages. At each reveal, ask whether the information was predictable, merely possible, or genuinely unavailable. Note the first update point. Judge execution, line quality, and process separately. This prevents one surprising card from rewriting the whole record.

Close the file with one next action

Write one conditional lesson that changes a visible behavior. It should tell you when to pause and what to compare. Schedule it for a small match set. Place every other observation in a parking list without turning it into another goal. The workshop is complete when the next action is clear, not when every turn has a label.

Record uncertainty honestly. If the replay did not show a needed detail or the opponent range cannot be reconstructed, mark the verdict unresolved. You can still retain a process lesson, such as noticing a constraint earlier, without claiming that one line was objectively best.

Delete or retain the replay according to your own privacy practice and platform terms. The field note should stand on minimal position facts and your reasoning. It should never require another player's identifier, a copied chat log, or a public accusation about their choices.