Evaluating the validity of a basketball performance metric in correlation with winning and losing games

Abstract

The aim of this dissertation project was to evaluate the validity of the novel Preventions metric, developed by Rodriguez (2013b), in correlation with performance indicators previously defined as discriminants of winning and losing games. The sample consisted of a selection of games from the 2012 Olympic Basketball Tournament for Women (n=18), categorised into two pools, close (n=9; x < 5 points) and unbalanced (n=9; 20 points < x < 40 points), based on final score-line differential. Box score data was scraped from the official FIBA Olympic archives to produce game-related statistics and advanced statistics; while Preventions metric data was recorded using hand notation. Descriptive statistics were generated to distinguish the differences in absolute values between the sample groups; Mann-Whitney U Tests were then performed to identify the significance of these differences. Winners and losers were differentiated by points prevented (p < 0.05), shots challenged (p < 0.05) and botched passes (p < 0.05). Pearson’s correlation data was produced to determine the strength of relationship between the new Preventions metric and academically proven advanced statistics. The variables correlating with the advanced statistics were points prevented (p < 0.01), shots challenged (p < 0.05) and botched passes (p < 0.05). The findings clearly identified points prevented, shots challenged and botched passes as significant variables to be considered as performance indictors due to their discriminant powers and correlation with peer-reviewed advanced statistics. This suggests that coaches should emphasise their defensive strategies regarding closing-out on shooters and disrupting passing lanes through on-ball pressure and off-ball denial defense.