Review trading rule violations with evidence

Assess outcome and adherence separately

A plan-following trade can win or lose, and a deviation can do either too. Use these four combinations in a review. This prevents a lucky deviation from automatically becoming a model to follow and a well-executed loss from automatically becoming a mistake.

Classification needs evidence

Record the rule, the version valid at trade time and the specific observation. If that information is missing, use “not assessable” instead of silently counting the execution as compliant. Unknown and compliant are different states.

For after-cost comparisons, both groups need the same period and a clear trade definition. Include group sizes and position sizes; two outliers should not appear to establish a stable rule effect.

Read the complete example

In the synthetic NQ sequence, six marked plan trades produce USD 170 net. Two marked deviations produce −USD 60, leaving USD 110 overall. This is a reconciliation of sample records, not evidence of a profitable future rule.