An Adaptive Horizon-Aware Model Selection Framework for Demand Forecasting under Horizon-Induced Degradation

arXiv:2602.13939v4 Announce Type: replace Abstract: Business environments characterized by intermittent demand, high variability, and multi-step planning horizons require forecasting policies that support consistent operational decisions across heterogeneous SKU portfolios. Because no forecasting model is universally dominant, and model rankings vary across evaluation metrics, demand regimes, and forecast horizons, selecting the most appropriate forecasting model for each demand series is a nontrivial model selection problem in inventory planning, procurement, and supply management. This study addresses this problem by proposing the Adaptive Hybrid Selector for Intermittency and Variability (AHSIV), a horizon-aware, structure-adaptive model selection framework designed to improve the consistency of forecast model assignment under horizon-induced ranking instability. The framework integrates the Metric Degradation by Forecast Horizon (MDFH), structural demand classification, multi-objective Pareto dominance, and hierarchical bias refinement within a unified decision-support scheme. The empirical evaluation is conducted on the Walmart, M3, M4, and M5 datasets under multiple train-test partition schemes and twelve-step forecasting horizons. The results show that AHSIV achieves statistical equivalence with the strongest single-metric baseline in aggregated performance while improving the consistency of model assignment across forecast horizons in heterogeneous demand settings. These findings suggest that forecasting model choice in multi-SKU environments should not be treated as a static ex post ranking exercise, but rather as a horizon-aware, structurally adaptive assignment problem that can provide more robust support for inventory planning, procurement alignment, and operational decision-making in business forecasting systems.