Artificial intelligence and soccer towards data-driven management of Moroccan soccer with exploratory clustering analysis of tactical and physical performance

Abstract

Abstract As Morocco prepares to host two major tournaments, the 2025 Africa Cup of Nations and the 2030 FIFA World Cup, this study investigates the contribution of artificial intelligence to tactical and physical performance analysis in Moroccan professional football. Using GPS data and Wyscout metrics, 15 matches were analyzed through PCA and clustering, producing three distinct performance profiles. Results show that winning matches were consistently associated with greater participation in high-intensity zones (Z4–Z5) and more explosive actions, while losses reflected reduced activity in these domains. On average, players covered 7,132 m per match, with clear variability across clusters. Principal Component Analysis explained 68% of the total variance (PC1 = 46%, PC2 = 22%), while clustering validation yielded a silhouette score of 0.67 and a Davies-Bouldin index of 0.42. These findings provide a data-driven framework that can support physical conditioning, tactical adaptation, and match preparation strategies in professional football. The paper provides a descriptive and exploratory modelling of the performance of a top-tier club (referred to as Club X) during the first half of the 2024/2025 Botola Pro D1 season. Results reveal distinct performance profiles across matches and tactical contexts, with three main clusters emerging: balanced contests, high-intensity encounters, and low-engagement games. These findings highlight the need for data-driven governance in Moroccan football, integrating workload indicators and tactical configurations. The study advocates implementing AI tools within coaching, scouting, and decision-making practices.