Identifying fast bowlers likely to play test cricket based on age-group performances

Abstract

A challenge for New Zealand cricket is identifying and retaining fast bowling talent. A methodology is described for determining individuals with a greater propensity to play test cricket for New Zealand, based solely on youth performances. Analyses were conducted on data from New Zealand youth test matches contested between 1986 and 2008, with subsequent selection for the test team as the dependent variable. Given the presence of collinearity and complex interactions amongst the variables, a regression tree technique was utilised, which ensured a pragmatic solution. After pruning, the regression tree indicated that average strike rate and average balls bowled per innings were key determining factors of future selection. Youth bowlers that had high workload (high average balls bowled) and high efficacy (low strike rate) accounted for 11/14 of those that went on to play test cricket for New Zealand. Interestingly, a further 2 of those 14 represented New Zealand in winter sports. Importantly, these results align with previous insights derived in literature. Extending this framework further enables the probability of playing test cricket for each individual to be determined by fitting a regression model to the regression tree residuals. This serves as a useful ranking system.