Measuring team performance and modelling the home advantage effect in cricket

Cricket differs from many team sports in that it is not generally played within easily definable constraints. Thus, it is difficult to compare and contrast team performances. By employing a range of linear and logistic modelling techniques this thesis quantifies the extent to which team quality effects and a range of associated factors such as home advantage have shaped team performance in Test, ODI and domestic cricket. With regards to the latter, the thesis established that, in all forms of cricket, a team's scoring potential and its capacity to win were both significantly amplified when it played at its home ground. The thesis proposes a method to estimate a projected score for the team batting second in ODI cricket. The method scales up the team's actual winning scores in proportion to its unused run scoring resources. This creates a projected victory margin when it wins with unused run scoring resources at its disposal and provides a more realistic measure of its relative superiority at the point of victory than the current wickets-in-hand method. Accordingly, the thesis recommends a revised scheme for recording victories in ODI cricket which is consistent across innings and provides a mechanism for all victories to be compared and ranked on an equal footing. The thesis employs linear modelling methods that account for the size of a victory in ODI cricket and the magnitude of the first innings lead in Test and domestic cricket to compute team ratings. The ratings are calculated independently of effects such as home advantage and quantify overall team performance relative to the average rating. They provide a robust measure of team quality and are not sensitive to the extraneous effects that may disproportionately impact on team performance. As a consequence, the thesis recommends that new methods be investigated to officially rate and rank teams in international cricket competitions. The team ratings also form the basis of a proposed outcome prediction model that can be instituted in Test cricket. The thesis established that a surprising trend has emerged in Test cricket, which confirmed that the team batting second, in general, has enjoyed a distinct winning advantage over its opposition. Accordingly, the thesis ascertained that relative strength during the final rather than penultimate innings significantly affected match outcomes and recommends that teams, when winning the coin toss, expose their strongest asset, whether this be batting or bowling, in the final innings.