Individual skill or just luck? : decomposing individual finishing and shot-stopping performance in European football

Expected goals (xG) are widely used to evaluate football performance. However, it is unclear whether players who consistently outperform their xG do so because of skill or because of luck. To our knowledge, this paper provides the first meta-analytic estimate of between-player heterogeneity in finishing and shot-stopping overperformance, extending existing team-level skillluck decompositions provided by Holzmeister & Johannesson (2025) to the individual player level. We analyse data from the English Premier League, La Liga, Bundesliga, Serie A, and Ligue 1 across eight seasons (2017/18-2024/25), totalling 3,202 outfield player-seasons and 932 goalkeeper-seasons. Using restricted maximum likelihood (REML) random-effects metaanalysis, we find that between-player heterogeneity accounts for 6.753-29.945% of total variance in finishing overperformance (average I² ≈ 21%) and 7.690-34.344% in shot-stopping (average I² ≈ 25%). Skill is detectable but modest. A complementary persistence regression shows that part of this overperformance carries over to the next season, both for outfield players (β = 0.110, p < 0.01) and goalkeepers (β = 0.185, p < 0.01). From these coefficients we derive that about 52% of a player's finishing skill signal is retained from one season to the next. For goalkeepers, the corresponding figure is roughly 74%. A single season of xG overperformance is too noisy to serve as a reliable indicator of player ability. Identifying genuine skill requires observations over multiple seasons.