1[2 {"id":"C1","text":"Ridge (lambda=50) RMSE is 36.5% lower than the naive forecast RMSE over 60 origins.","results_keys":["rmse_ridge","rmse_naive","rel_improvement_vs_naive"],"status":"supported","caveat":"Current-vintage data, publication lags ignored."},3 {"id":"C2","text":"Ridge does not beat the expanding-mean baseline: its RMSE is 3.1% higher and its MAE also higher.","results_keys":["rmse_ridge","rmse_mean","rel_improvement_vs_mean","mae_ridge","mae_mean"],"status":"supported","caveat":"Single 60-month window; fixed lambda."},4 {"id":"C3","text":"The ridge-vs-naive gain is statistically distinguishable from zero (DM p=0.0014), while the ridge-vs-mean difference is not (DM p=0.59).","results_keys":["supplementary"],"status":"supported","caveat":"Supplementary, simple DM test with t reference on 60 errors; not required by the task."},5 {"id":"C4","text":"All standardisation statistics are computed from each origin's training window only.","results_keys":["normalisation_window","n_analysed"],"status":"supported","caveat":"Verified by a future-data perturbation test recorded in out/state/checks.md."}6]7