nx_flip_hazard.nx

nx_flip_hazard.nx -- THE WEAR-AND-TEAR MEMBER of the FlipOrgan family. nx_flip_hazard <class> <comp> <usage_now> <hold> <repair_cost> [plane-prefix-root] WHY IT EXISTS (measured on the LIVE engine 2026-08-08, not assumed): nx_flip_score decides with `margin = lift - repair`, verified exactly across all 8 parts of the FOCUS13A deal, and the findings plane's `cond` column appears in ZERO decisions. So a part the engine declines to fix carries a RISK OF FAILING that is priced at ZERO. For the DPS6 dual-clutch that unpriced risk is 29.3%-51.9% of the entire p10 downside margin. It reads observations ITSELF from <root>obs- (id|class|comp|usage|source) -- the caller never types a failure rate, exactly as nx_flip_score reads parts rather than being handed a repair total. Thresholds come from <root>hzcfg- (class|min_obs|min_r2_bps|note), never from code. ASSET-AGNOSTIC: `usage` is miles for a car, trading days for an FX drawdown, cycles for a battery. PROVEN on both -- DPS6 beta=1.7274 (WEAR-OUT) vs EUR/USD drawdowns beta=0.6226 (PERSISTENT). Opposite regimes from one estimator; car logic applied to FX would be backwards. PROVENANCE CEILING, printed on every run because it bounds every number downstream: NHTSA complaints are SELF-REPORTED with no exposure denominator, so this is the hazard of COMPLAINT ARRIVAL, not of failure. Real evidence about shape and ordering; NOT a fleet-calibrated census. license_tier: ORIGINAL No hw writes (Rule 26). expect_exit: 0