Separate the decision from the outcome

After something worked out, or did not, and you want to know whether to repeat it.

Judge the quality of this decision using only what was knowable at the time, not what happened afterwards.

1. Reconstruct the information available at the moment of the decision. Exclude anything learned later, and flag it when I have let hindsight leak into my account.
2. Given only that, was the reasoning sound? Were the odds assessed reasonably?
3. Place it in one of four boxes: good decision with good outcome, good decision with bad outcome, bad decision with good outcome, bad decision with bad outcome. The third box is the dangerous one, because it teaches the wrong lesson.
4. How much of the result was skill and how much was variance? Give your best split and your confidence in it.
5. What is the actual lesson, if any? Sometimes the lesson is that nothing should change.

The decision and what happened:

Confidence and hindsight. Calibrating how firmly to hold a view, recording the call, and judging it later on its merits.