Retail inventory decisions often rely on managerial intuition or static rules even when rich demand forecasts and contextual data are available. This study applies the single-period newsvendor model to a retail dataset comprising 73,100 product-store-day observations containing demand forecasts, pricing, promotions, weather, and competitor information. We construct an empirical demand distribution using a forecast-error framework supported by highly accurate forecasts (correlation ≈ 0.997 with actual demand). Procurement, holding, and shortage costs are estimated using retail-appropriate assumptions, enabling calculation of the critical fractile and optimal daily stock levels. Results reveal a substantial gap between cost-optimal recommendations and actual retailer behavior: the retailer maintains near-zero stock-outs (≈0.7%) while the model-optimal policies would reduce inventory by 60-70% achieving approximately 48% cost savings. This discrepancy suggests a loss-averse newsvendor behavior where retailers implicitly prioritize service level over cost efficiency. The findings demonstrate how classical models benefit from modern forecast-based estimation and highlight opportunities for reducing costs through calibrated shortage cost assumptions and multi-period inventory strategies.