How much to stake on each proven edge. Once Calibration shows the odds are honest, the next question is size. For every measured cohort this reads its own settled record — realized win rate, average win vs average loss — and returns the fraction of a risk budget the Kelly criterion justifies, plus the expected edge. We show full Kelly but recommend half, cap any single call, and size nothing whose downside we haven't measured. (This is the across-the-edges view — for per-share sizing on one ticker, use QL Position Sizer.)