A Computational Exploration of Football Player Performance at the FIFA World Cup 2022 Using StatsBomb Event and 360° Spatial Data

Abstract

Study purpose. This study explores the descriptive and structural characteristics of individual player performance at the FIFA World Cup 2022 using rich event and 360° spatial data, with the aim of demonstrating that advanced metrics provide a more discriminative profile of player contribution than conventional box-score statistics.

Materials and methods. Open event data and 360° freeze-frame data for all 64 matches were extracted using the statsbombpy interface and processed through a reproducible Python pipeline adapting the CRISP-DM workflow. Twenty-five advanced per-90-minute metrics were engineered for 681 players, of whom 346 met a minimum-playing-time threshold of 180 minutes. Descriptive statistics, Pearson and Spearman correlations, and K-means clustering were applied.

Results. Expected goals correlated strongly with goals scored (r = 0.75), whereas expected assists correlated only weakly with recorded assists (r = 0.22), revealing a substantial pool of creative contribution that conventional statistics fail to capture. Position and confederation profiles differed systematically, and unsupervised clustering recovered interpretable playing-style groups that cut across formal positional labels.

Conclusions. Advanced event and spatial metrics expose tactically important, under-recognised contributions and provide a reproducible analytical baseline for subsequent network- and graph-based player evaluation