Modeling team performance

This study of 3366 Bundesliga-match results aims at a considerable extension of previous analyses of football teams’ performance on the playground. On the one hand, this extension is rooted in a comprehensive consideration of theoretical aspects that, so far, have been neglected in the literature on football teams: Team composition and institutionalization processes. The empirical results indicate that teams that are heterogeneous on several relevant traits perform significantly worse than more homogeneous teams. Moreover, even though the Flat Back lineup (“Abwehrkette”) became tremendously popular within the 1992–2003 observation period—thereby representing a prime example of an institutionalized practice—substantial regression analyses do not indicate any performance advantages of this innovative defense system. Thus, we are unable to reject the new institutionalism’s view that novel organizational procedures might successfully diffuse without possessing any practical added value for the adopting organizations! On the other hand, we provide a novel suggestion to control for regression-to-the-mean-phenomena in statistical models of football match results: Team ranking covariates prove as elegant and powerful controls for regression-to-the-mean effects within the fixed effects regression setup.