AI-Driven Demand Forecasting for Sporting-Goods and Teamwear Supply Chains Under Seasonal and Tournament Demand

Abstract

This study addresses the persistent forecasting problem faced by sporting-goods and teamwear supply chains where seasonal cycles, tournament schedules, team performance, promotional activity, short selling windows, product variety, and rapidly changing customer preferences can create substantial demand uncertainty, stockouts, excess inventory, missed sales, and inefficient replenishment. The purpose of the study was to quantitatively examine how AI-driven demand forecasting capabilities influence demand forecasting and supply-chain performance under seasonal and tournament demand conditions.The study retained 238 valid responses from demand and forecast planners, supply-chain managers and analysts, inventory and replenishment managers, procurement professionals, merchandising and sales-planning personnel, and operations or logistics managers selected through purposive sampling. The key independent variables were AI-based historical demand pattern learning, real-time demand data integration, seasonal and tournament demand sensing, adaptive forecast updating capability, and human-AI forecasting integration, while demand forecasting and supply-chain performance served as the dependent construct. Data were collected using a structured five-point Likert questionnaire and analyzed through descriptive statistics, Cronbach's alpha, KMO and Bartlett's tests, Pearson correlation, multiple linear regression, and regression diagnostics in SPSS. Seasonal and tournament demand sensing recorded the highest mean (M = 4.18, SD = 0.57), while overall demand forecasting and supply-chain performance reached M = 4.09 (SD = 0.55). All five capabilities were positively correlated with performance, with tournament demand sensing showing the strongest relationship (r = .69, p < .001). The regression model explained 62.9% of performance variance, R² = .629, adjusted R² = .621, F (5,232) = 78.43, p < .001. Tournament sensing was the strongest predictor (β = .31), followed by historical demand learning (β = .24), real-time integration (β = .19), adaptive updating (β = .16), and human-AI integration (β = .13). These findings imply that event-sensitive AI forecasting, supported by timely data integration, adaptive updating, and structured human judgment, can strengthen forecast accuracy, inventory alignment, product availability, replenishment responsiveness, and supply-chain planning effectiveness.