1[2 {"id": "C1", "text": "Over 60 origins (2021-09 to 2026-08, h=1), the ridge forecast has lower RMSE than the naive forecast, a relative improvement of about 36%.", "results_keys": ["rmse_ridge", "rmse_naive", "rel_improvement_vs_naive", "n_origins"], "status": "supported", "caveat": "Current-vintage data, publication lags ignored."},3 {"id": "C2", "text": "The ridge forecast does not beat the expanding historical mean: its RMSE is about 3% higher and its MAE is also higher.", "results_keys": ["rmse_ridge", "rmse_mean", "mae_ridge", "mae_mean", "rel_improvement_vs_mean"], "status": "supported", "caveat": "Fixed lambda=50, not tuned."},4 {"id": "C3", "text": "The ridge advantage over naive is statistically significant (DM p about 0.001); the difference between ridge and the expanding mean is not (DM p about 0.59).", "results_keys": ["dm_stat_ridge_vs_naive", "dm_pvalue_ridge_vs_naive", "dm_stat_ridge_vs_mean", "dm_pvalue_ridge_vs_mean"], "status": "supported", "caveat": "Only 60 evaluation points; DM assumes stationary loss differential."},5 {"id": "C4", "text": "All standardisation statistics were computed only from each origin's training window (expanding), with 413 training rows at the last origin.", "results_keys": ["normalisation_window", "n_analysed"], "status": "supported", "caveat": "Verified by future-data corruption test in out/state/checks.md."}6]7