Each row is a cohort — a slice of your trade history filtered by the selected dimension.
HIT %: percentage of wins. Above 55% is good. Below 45% is a leak.
EXP: expectancy in R. Positive means you make money on average; negative means you bleed.
AVG CONF: what the model thought of these trades on average. If high conf but low hit rate → the model is wrong here, postmortem time. If low conf but high hit rate → your discretion is beating the model, build training data for this pattern.