Live track record. Average forward return of published screens minus the benchmark — 1 week out: +0.90% vs the Russell 2000 Growth ETF (9 frozen cohort readings / 2 independent); -0.13% against 100 randomly chosen eligible names, equally weighted like the screen itself. Each published screen’s return is measured once, about a week after publication, then frozen — never recomputed and never backfilled — so readings accumulate as real forward evidence. Cohorts published within the same week overlap, so the independent count, not the total, is the honest sample size. Two yardsticks, deliberately. The index tells you whether the whole package beat simply buying small-cap growth. The random draw tells you something narrower and more important: whether the ranking is doing anything, with the universe and the equal weighting held constant. Equal-weighting a small-cap universe has historically paid something on its own, so a lead over the index alone could be the weighting rather than the picking. The draw is seeded from the date, so it cannot be re-rolled until it flatters.