Application of social network analysis in football match analysis: A systematic review

Abstract

Social network analysis (SNA) has demonstrated strong potential for application in football match analysis. Previous reviews have limitations such as insufficient football-specific focus, lack of analysis-type classification, and outdated coverage. This review provides a comprehensive and up-to-date synthesis of the literature, categorizing applications of SNA in football matches into descriptive, correlational, comparative, and predictive analytical types. Following PRISMA 2020 guidelines, articles from the Web of Science All Databases and Scopus were retrieved using targeted keyword combinations. Of 1208 identified records, 49 articles satisfied inclusion criteria and were fully reviewed. Social network descriptive analysis primarily focused on identifying key players and their interactions. Correlational studies examined associations between network metrics and match performance indicators. Physical demands showed moderate, position-dependent correlations with degree-based centrality metrics. Technical indicators exhibited small to moderate positive correlations with network metrics. Match outcomes were weakly associated with structural metrics such as total links, network density, often serving as situational factors. Most studies (51%) employed comparative analysis. Micro-level analysis compared centrality metrics to identify key players and their attributes across varying situational factors, with midfielders exhibiting the highest centrality. At the macro level, studies comparing metrics such as network density found that superior network properties are linked to better team performance and vary across situational factors. Predictive studies demonstrated that network metrics possessed significant predictive potential, and models that incorporated these metrics achieved superior performance. Overall, social network predictive analysis accounts for the smallest proportion (12%). The predictive potential of SNA remains underexplored and warrants further scholarly attention.

PROSPERO registration number: CRD42024587155