1# Verification checks (P4)2| check | outcome |3|---|---|4| Clean re-run (results.json, forecasts.csv deleted, regenerated twice) byte-identical | PASS |5| forecasts.csv has 144 rows: 36 per (model, horizon) | PASS |6| target_date − origin_date = horizon months for every row | PASS |7| Naive forecasts independently recomputed as last non-missing obs <= origin | PASS |8| Exactly 1 missing actual per (model,horizon) (target 2025-10-01) -> n_evaluated = 35 each, 36−1 | PASS |9| MAE <= RMSE for every model/horizon; RMSEs > 0 | PASS |10| rel_improvement = 1 − rmse_model/rmse_naive (computed directly in code) | PASS |11| Chronology: first_origin 2023-07-01 < last_origin 2026-06-01; training data always truncated to <= origin (s.index <= t in code) | PASS |12| Exclusion ledger: 945 − 1 − 3 = 941 = n_analysed | PASS |13| Magnitude vs baseline: model forecasts differ from naive by at most 0.016 pp; model RMSE within 3% of naive | PASS (plausible: AR(2)-diff coefficients small, intercept ~0) |14