Abstract:

Accurate forecasts of football game results are interested not only by statisticians due to the peculiar nature of the data but also by football bettors for financial purposes. The aim of the study is to predict outcome of individual football (soccer) games between two teams based on their past performance using two different data mining techniques: Artificial Neural Networks and Decision trees. Windraw- lose match results are forecasted by using the overall performance of teams and game outcomes played between the two teams in the past seasons. Game data are gathered from three leagues (Turkish Super League, English Premier League and Brazil Serie A) selected from different regions of the world. Accuracy of forecasts obtained from artificial neural networks and decision tree techniques are compared. Model out-comes from three leagues are also compared to understand country wise behavioral differences of football games.