Define the setup before analyzing it
Describe the entry condition, valid time window and an exclusion criterion. “VWAP rejection” alone leaves distance and confirmation undefined. Preserve earlier definitions when changing the rule later; otherwise different rules become mixed under one tag.
Include sample size and distribution
Two winning trades on Tuesday do not establish a Tuesday effect. Show trade count, costs, average size and the period for each group. Check whether one outlier dominates the mean. A colored heatmap displays those records; it does not add independent evidence.
Start with a few groups justified in advance. Testing many times, weekdays and filters retrospectively makes chance winners easy to find. Check a discovered hypothesis on a later period with its definition unchanged.
An example of a focused question
Question: do planned and late entries for the same setup differ after costs? Keep market and session consistent and classify lateness using a predefined rule. Outcome, adherence and sample size remain three separate parts of the answer.