1[2 {"id":"C1","text":"The seasonal AR model has lower out-of-sample RMSE than the seasonal-naive forecast at h=1, 2 and 4 (relative improvements 0.23, 0.13, 0.19).","results_keys":["rmse_model_h1","rmse_snaive_h1","rmse_model_h2","rmse_snaive_h2","rmse_model_h4","rmse_snaive_h4","rel_improvement_rmse_h1","rel_improvement_rmse_h2","rel_improvement_rmse_h4"],"status":"supported","caveat":"Point estimates over 40 origins; losses dominated by the 2021-2023 surge."},3 {"id":"C2","text":"The model also has lower MAE than seasonal naive at all three horizons.","results_keys":["mae_model_h1","mae_snaive_h1","mae_model_h2","mae_snaive_h2","mae_model_h4","mae_snaive_h4"],"status":"supported","caveat":"Descriptive only."},4 {"id":"C3","text":"At h=4 the mean squared-error loss differential (2.82) is not statistically distinguishable from zero using a Bartlett HAC SE with lag 3 (DM=1.35, p=0.18).","results_keys":["loss_diff_mean_h4","loss_diff_se_h4","hac_lags_h4","hac_kernel_h4","dm_stat_h4","dm_pvalue_h4"],"status":"supported","caveat":"Conclusion is sensitive to bandwidth: Bartlett lag 6 gives p=0.04 (post-hoc sensitivity); HLN correction gives p=0.23."}5]6