The sacking illusion: A counterfactual analysis of mid‑season coaching changes using points and expected points in European football

Mid-season coaching dismissals are common in professional football, yet it remains unclear whether they improve team performance or merely reflect regression to the mean. We analysed 331 mid-season coaching changes in the first and second men's divisions of the five highest-ranked European countries (2017/18-2021/22) using match-level data. Teams that dismissed their coach were matched to control teams with identical five-match performance trajectories that retained theirs, and actual points (P) and expected points (xP) were compared over the ten matches following each dismissal. Prior to dismissal, P collapsed from 1.28 to 0.34 points per match whereas xP remained stable (1.26-1.38), a pattern almost identical in treated and control teams. Following the coaching change, no significant treatment effects were observed: average treatment effect on the treated (ATT) estimates ranged from -0.18 to +0.13 for P and from -0.06 to +0.05 for xP, with all confidence intervals including zero. Mid-season coaching changes thus produced no detectable improvement relative to matched controls. The divergence between collapsing results and stable underlying performance suggests that clubs react to outcome variance rather than genuine decline. Integrating expected goals-based trajectory analysis into coach evaluation could help clubs avoid unnecessary and costly dismissals.