1[2 {"id":"C1","text":"Ridge (lambda=50, 8 standardised features) has lower out-of-sample RMSE than the naive forecast, a 36.5% RMSE reduction, statistically significant by Diebold-Mariano (p≈0.001).","results_keys":["rmse_ridge","rmse_naive","rel_improvement_vs_naive","dm_stat_ridge_vs_naive","dm_pvalue_ridge_vs_naive"],"status":"supported","caveat":"60 origins; current-vintage data, publication lags ignored."},3 {"id":"C2","text":"Ridge does not beat the expanding-mean baseline: its RMSE is about 3.1% higher (rel. improvement -0.031) and MAE is also higher.","results_keys":["rmse_ridge","rmse_mean","mae_ridge","mae_mean","rel_improvement_vs_mean"],"status":"supported","caveat":"Specific 2021-09..2026-08 evaluation window and fixed lambda."},4 {"id":"C3","text":"The difference between ridge and expanding mean is not statistically significant (DM p≈0.59).","results_keys":["dm_stat_ridge_vs_mean","dm_pvalue_ridge_vs_mean"],"status":"supported","caveat":"DM test approximate for nested estimated models; small sample."},5 {"id":"C4","text":"Standardisation statistics were computed from each origin's training window only (expanding), with 413 training rows at the last origin.","results_keys":["normalisation_window","n_analysed"],"status":"supported","caveat":"Verified by leakage perturbation test in out/state/checks.md."}6]7