Foundation 03Review preview

A good result cannot repair a weak decision.

Poker outcomes contain variance. Review the action using what was knowable when it was chosen, then use the result as new evidence—not as a verdict that rewrites the original reasoning.

10–12 minutes3 judgment checksdecision-quality-content-v1

Review model

Freeze the decision before reading the outcome.

1. Reconstruct: position, stack, pot, cards, action history, and legal options.
2. Record: your range assumptions, intended plan, chosen action, and confidence before feedback.
3. Evaluate: compare the reasoning with an appropriately labeled rules, math, benchmark, or solver source.
4. Observe: keep the result, but do not let one result stand in for expected value or decision quality.

Evidence boundary

Confidence is a calibration signal, not a score.

Retrieval-practice research supports answering before feedback, and calibration research motivates recording confidence. Neither proves that this exact poker lesson improves skill. High confidence can be accurately calibrated, underconfident, or dangerously wrong; it must be compared with correctness over enough attempts.

Active recall

Protect the original decision from hindsight.

Your first checked answer counts toward this session. No answer or reflection is persisted.

Before feedback, define a fair review.Confidence before feedback

1. A carefully reasoned action loses this hand. What follows?

The action was necessarily bad The outcome is negative; decision quality still requires a decision-time review The action was necessarily good because it was carefully explained

2. What does a 5/5-confidence wrong answer provide?

A high skill score A calibration signal to compare across repeated attempts A reason to ignore confidence entirely

3. What should a disciplined review inspect first?

The winner and final pot only The decision-time state, assumptions, legal actions, reason, confidence, and versions How the result felt afterward

Freeze the decision before reading the outcome.

Confidence is a calibration signal, not a score.

Protect the original decision from hindsight.

The decision stays frozen; the result stays separate.