Akademik Veri Yönetim Sistemi
VII. International Applied Statistics Congress (UYIK-2026), İstanbul, Türkiye, 11 - 13 Mayıs 2026, ss.1, (Özet Bildiri)
The expected goals (xG) metric and passes per defensive action (PPDA) are widely used in football analytics, yet their joint contribution to team success and the heterogeneity of these effects across leagues remain underexplored. This study examines whether xG outperforms actual goals as a predictor of next-season points, whether pressing intensity influences points directly or through xG as a mediator, and whether these relationships differ across the top five European leagues. Season-aggregated and match-level data from Understat covering 2014 through 2019 across the English Premier League, La Liga, Bundesliga, Serie A, and Ligue 1 were analysed (588 team-season observations, 21,700 team-match records). Pearson correlations, paired bootstrap tests, multilevel linear models with random intercepts and slopes, and bootstrap mediation analyses were applied. xG and actual goals showed comparable next-season predictive accuracy (Δr=+0.012, 95%CI [−0.018,+0.041], p=0.43), challenging the common claim that xG is a markedly stronger forward-looking indicator at the seasonal scale. Multilevel models indicated a strong association between xG and points (β=0.36, p<0.001) and a smaller but reliable association between pressing intensity and points (β=0.04, p=0.002), with no statistically detectable variation in slopes across leagues (LR χ²=7.78, p=0.17). Mediation analysis showed that approximately 75% of the total effect of pressing on points was transmitted through xG, with the proportion mediated ranging from 63% in the Bundesliga to 93% in La Liga. A home-and-away robustness check confirmed that pressing remained relevant in both contexts, with a slightly larger coefficient in away matches. Results support a chance-quality channel through which pressing translates into points and reveal league-specific tactical signatures that aggregate league-table outcomes obscure.