Prediction of in-play tennis

Tennis is one of the most volatile sports, as the favourite to win the match can change within a couple of service games. The growth of technology over the past decade has also resulted in an exponential increase of in-play wagering. Thus, the aim of this thesis was to investigate the game of tennis whilst in-play in order to increase the accuracy of predictions by updating the probability of winning a point on serve using various approaches. Analysis was performed to determine whether incorporating in-play serving and winning statistics, court-side dominance and point importance provided an accurate representation of match progression. Analysis on the efficiency of the in-play wagering markets was also performed and a method to incorporate over-round into a Markov Chain model was establish.