According to behavioral economics, coaches may be unconsciously biased, and this could lead to deviations from rational behavior, which in turn affects team performance. We analyze the influence of a particular behavioral bias of coaches, overconfidence, on the performance of soccer teams. We use a sample of 63 coaches managing all the soccer clubs involved in the Russian Football Premier League during the four seasons between 2010 and 2013/2014. To measure overconfidence, we use a press-based metric that is generally accepted in corporate governance studies and complement it with an additional continuous measure. Coaches' overconfidence positively and significantly influences team average scores, both in the baseline regression and robustness checks. Additional testing allows us to draw conclusions regarding the inverse U-shaped relationship between overconfidence and performance. We cannot conclude that overconfidence has any effect on coaches' risk-taking that can be approximated by goals scored or allowed. We apply the well-studied methodology of overconfidence measurement to the new field of sport economics, thereby generating novel results. Although overconfidence is perceived negatively in corporate governance, we show that in sport, it is beneficial to be overconfident. The findings contribute to sport literature, more specifically to the field of performance in soccer, with results that support the importance of a coach's personal traits.