An applied framework for multi-domain player evaluation in professional football: Ranking agreement between traditional and expected-goals-derived metrics.

Abstract

This study proposed an applied framework for evaluating football player performance using expected-goals (xG)-derived metrics and examined how player rankings from these metrics differed from those produced by traditional event-based indicators. Match-event and tracking-derived data from the 2025 K League 1 season were analysed across four domains: finishing, shooting, chance creation, and goalkeeping shot-stopping. Ranking agreement was quantified using Spearman's ρ and Kendall's τ, the mean absolute rank difference (also expressed as a percentage of N), and the proportion of players exceeding predefined absolute (≥5 and ≥10 positions) and relative (≥20% of N) rank-shift thresholds interpreted as applied heuristics rather than validated cut-offs. Moderate-to-strong correlations were observed across all four domains (ρ = 0.58 to 0.85, all p < 0.001); however, substantial individual-level discrepancies persisted, with mean absolute rank differences of 3 to 16 positions and at least 44% of outfield players re-ranked by ten or more positions. Because the study does not validate either ranking against an external criterion, it establishes that traditional and xG-derived metrics are not interchangeable rather than that either ordering is more valid. xG-derived metrics are best interpreted as complementary, context-sensitive information alongside traditional indicators for scouting, recruitment, and tactical decision-making.