Prediction of Swedish Harness Racing.

Abstract: Harness racing is a sport where betting most often is done based on historical performances and known conditions from each horse. With up to 12-15 horses in each race and with a quite large set of data for each horse, harness racing seemed to be very suitable for some statistical modeling and regression analysis. The main goal of this project was to construct a model that predicts the outcome of a race better than the odds. To achieve this, many dierent covariates, and combinations of them, have been tested. Also dierent types of regression methods, such as logistic regression have been tested in order to find the best model. A big challenge has been to collect a very large amount of useful data and handling it in an e-cient way. In the end several models had been developed whereof the best ones made better predictions than the odds. Also a few betting strategies have been developed in order to investigate the possibility of making money by using the models. At least one of them seem to provide a good return.