Predicting the winner of NFL-games using Machine and Deep Learning

Predicting the winner of American Football games has been a hot topic for decades. Since the 1950’s people have tried to classify the strength of an athlete or team and predict the outcome of sports games. More recently, people have tried to predict the binary outcome of a game-winner or to estimate the number of goals applying Machine Learning methods. In recent past Deep Learning methods have been widely developed and have a significant impact in research fields such as Natural Language Processing and in 2015 Google has opened the open-source software library Tensorflow which enables users to perform Deep Learning methods in a more efficient way. The increase of open-source data of NFL-games and the availability of open-source software for applying Deep Learning methods makes one wonder if Deep Learning methods can outperform classical Machine Learning methods in predicting the winner of NFL-games. In this paper we analyze if Deep Learning can outperform Machine Learning in predicting the winner of NFL-games.