Cricket performance measures

Abstract

The purpose of this paper is to present a summary of batting and bowling performance measures that have been developed over a period of eight years. The measure of batting performance (BP) consists of three components. Firstly, an exponentially weighted average (EWA) is calculated in which recent scores have higher weights than scores further back in time in order to get a measure that adequately reflects the batsman’s present form. EWA is then adjusted up- or downwards depending on the batsman’s consistency and strike rate. This measure was subsequently generalized to BPW which includes weights for runs scored against specific countries in home or away matches separately. In the case of bowling, the first challenge was to develop a single comprehensive measure of bowling performance, the combined bowling rate (CBR) for limited overs matches and the dynamic bowling rate (DBR) for unlimited overs matches. Hence, by alsotaking bowling consistency into account, the current bowling performance measure (CBP) wasdeveloped. This measure was also extended to include weights for wickets taken and runs concededagainst each international team in home or away matches. These measures contain parametersthat were based on players’ scores up to the specific date of calculation. Before using the latestavailable data for the present presentation, it is logical to update these parameters. This hasbeen done in the present study. If a player performs well at local level, it is important to be ableto compare his performance with those of players already in the national team. Weights havebeen determined for runs scored by batsmen and also for runs conceded and wickets taken bybowlers in local matches, whereby the performances of players can be compared irrespective ofthe proportion of international matches played by each player. After having used the measures onvarious occasions the need arose to revise some of them slightly. The updated measures are thenused to give rankings of South African batsmen and bowlers in First Class and List A matchesalike. Methods have also been developed to compare the batting and bowling performances ofplayers after a short series of matches. It was shown that the traditional average could haveunrealistic values in the case of a batsman who had a large percentage of not out scores in ashort series. Alternative measures have been developed and shown to give much more realisticvalues. In the case of bowling it is important not only to count the number of wickets a bowlerhas taken, but to distinguish between the wickets of top and middle order batsmen on the onehand and lower order batsmen on the other. A system has been established whereby weights areattached to the wickets according to the batting position of the batsman. Instead of just countingthe number of wickets taken, the sum of the weights of the wickets taken by the bowler is usedin the calculation of CBR* or DBR*. Finally, the latest form of the game, Twenty20 cricket, isalso discussed.