Using neural networks to predict binary outcomes

Predicting binary outcomes with respect to sports typically involves the use of human judgement and subjectivity. This research considers the use of neural networks and logistic regression as an interesting and quantifiable alternative to human tipsters, whose predictions are of a qualitative nature. Using a set of naive but objective input variables, we highlight the improvement in prediction accuracy obtained through the use of neural networks and logistic regression. A comparison between neural networks, logistic regression and human tipping experts is presented.