Successful applications of statistical modeling to betting markets

摘要

With a strategic theme of entrepreneurship, Swinburne University encourages innovation and commercialisation of research results. Swinburne Sports Statistics initially embraced this theme by selling computer predictions of Australian rules football to the media, and since 1998 predictions for a range of sporting events were placed on our web site,. The expansion of sports betting created a market for sports prediction expertise, and we discuss profitable attempts at demonstrating the inefficiency of football, cricket and rugby betting markets. The head to head nature of most sports betting results in low margins for bookmakers, and the lengthy time many events take to complete also results in higher risk and lower turnover. To overcome these drawbacks, 'Betting in Running' allows betting on events which occur within a sporting contest, such as the number of runs in an over of cricket, or the point score in a single game of tennis. Sportsbet21, a Swinburne startup company, provides to bookmakers computer generated odds driven by a statistical model. Models have been developed for both cricket and tennis, and operated successfully by Ladbrokes in the UK for some time. The product has achieved profit targets on growing turnover, and demonstrated the robustness of the mathematical models.