Judging decisions rather than results

A good bet can lose and a bad one can win, and over any short run both happen constantly. Reviewing by result trains you toward whatever was lucky recently.

The result of a single bet contains one bit of information heavily contaminated by variance. The reasoning contains far more, and it is available immediately rather than after hundreds of repetitions.

Results are not irrelevant. Over hundreds of bets they are the only measure that matters, which is why the horizon of the judgement is the whole question.

And that is the trap. Outcome-based review feels rigorous because it uses real data. It is the most common route to abandoning a working process and adopting a broken one.

What to do with it. Grade each bet's reasoning at the time and the results only in aggregate, at a horizon measured in hundreds.