Integrating sentiment analysis into football scouting: A data-driven approach

Football scouting has undergone a significant transformation with the integration of data-driven methods. However, existing approaches focus primarily on quantitative performance metrics, while overlooking nuanced insights from qualitative data such as text data. Therefore, this study explores the potential of sentiment analysis in football scouting by analyzing fan comments on social media platforms to evaluate player perception. A manually labeled balanced dataset comprising 750 comments in four languages (English, German, French, and Spanish) was collected and processed using three sentiment analysis models: VADER, XLM-T, and GPT-4o-mini. A comparative evaluation based on accuracy, precision, recall, and F1-score demonstrated that GPT-4o-mini outperformed the other models, achieving an accuracy of 76%. To translate these insights into practical value, a user-friendly dashboard was developed, allowing scouts to visualize sentiment trends and player-specific feedback through intuitive charts and sample comment displays. To gather initial practitioner perspectives, our dashboard was presented to the scouting department of VfB Stuttgart 1893 AG, where exploratory expert feedback provided early indications of how sentiment analysis might complement player evaluations and transfer decisions. The results indicate the potential value of combining quantitative and qualitative insights in modern football scouting, suggesting that sentiment analysis could potentially complement decision-making in player recruitment.