The topic of build up play in football is more relevant than ever before. We present a way of modelling and learning about build up play through machine learning. Utilising clustering as a way of describing defensive lines and using logistic regression to predict the probability of the attacking team playing through the lines. The results examine how a high first press influences the attacking teams opportunities. Furthermore we look at the importance of starting the buildup in a central position and how that relates to the outcome of the sequence of play. The thesis also presents a way of successfully combining proposed models with related works. We find that our models are a viable way of analysing build up play and can be used as a means for answering football specific questions.