This thesis analyzes the prediction accuracy of the wisdom of crowds on the outcomes of football matches. A simple prediction model, based on player valuations reported on the football statistics website ‘’, is drafted and used to predict the league results of the Premier League, Bundesliga and Primera División over the period 2013-2016. The resulting prediction accuracy is compared with that of four benchmark methods: ELO ratings, aggregated betting odds, a ‘home team wins’ model and pure chance. My analysis shows that the wisdom of crowds model is a better prediction method than pure chance and the ‘home team wins’ model, is able to compete with ELO ratings, but is clearly a worse forecasting method than the model based on aggregated betting odds. Furthermore, I investigated the presence of biases that potentially could influence the valuations of users on ‘’. I find that cognitive biases are likely to influence user responses and therefore could undermine the prediction accuracy of the corresponding model. This could explain why the wisdom of crowds does not outperform aggregated betting odds. Other factors might distort the validity of user stated player valuations; further research should therefore try and focus on examining the impact of these confounders on the valuations of Transfermarkt users.