This paper studies a simple game theoretical model to learn whether we can rationalize the observed defensive behavior of teams that have the objective of "point-maximization". The data collected is from the nine Champions League seasons that took place between 1997 and 2006. From the data we obtain an implicit measure of team's strategy, and estimate how these strategies affect game outcomes. The richness of the dataset allows us to include dynamics and team asymmetry in the analysis. Using our estimates, we can find the optimal strategy for each team in different stages and fields of the soccer match.

The measure for the team’s strategy, as well as the model, are successful replicating the empirical findings that asymmetry is important (home teams attack more than when playing away; weaker teams defend more than stronger ones), and also half-time results play a role in second half strategies (teams ahead at the break defend more than when behind; the magnitude of that result also matters). Our main finding is that, given our estimates and the “point- maximization” objective, teams should play more offensively than what the observed data suggests. Hence, it is difficult to rationalize overly defensive behavior with this model. Rational point-maximizing teams should only choose to adopt a defensive strategy when one team has significant home-field and/or quality advantage and if the weaker team is leading the half-time score by a significant large difference.