From player actions to player profiles: Development and validation of a context-adjusted, position-specific Player Stat Index in elite European football

Player statistics are used to evaluate and recruit footballers, but the same action can indicate quality, tactical role, or team exposure. We developed a context-adjusted, position-specific Player Stat Index (PSI) using 1751 matches from the 2025/26 Premier League, La Liga, Bundesliga, Serie A, and Ligue 1. Heatmaps separated players into 15 spatial roles. Action volumes, attempt-shrunk efficiencies, and player shares of team production were adjusted for possession, chance environment, opposition, recent team strength, home advantage, and league. All parameters were frozen before a chronological 25% test period. Prior fine role predicted the next observed role in 67.0% of 12,115 test player-matches, versus 41.1% for a broad-position baseline. Buildup (Spearman rho = 0.439, 95% CI 0.409–0.469), ball security (rho = 0.395, 0.365–0.422), and creation/progression (rho = 0.334, 0.307–0.360) showed the strongest domain persistence. PSI had a small association with next-match rating (rho = 0.152, 0.124–0.177). In held-out actual-XI models, adding PSI to basic context increased win AUC from 0.543 to 0.573; adding PSI to recent xG and possession increased AUC from 0.584 to 0.602 (delta = 0.018, −0.020 to 0.056). Removing average win/draw effects from historical performances produced highly similar rankings (rho = 0.981) and almost identical validation. PSI provides limited prediction of subsequent performance and modest team-outcome information, but its strongest use is as a multidimensional, within-role profile for player evaluation and recruitment.