Comparing the performance of baseball players: a multiple output approach

This article extends ideas from the economics literature on multiple output production and efficiency to develop methods for comparing baseball players that take into account the many dimensions to batting performance. A key part of this approach is the output aggregator. The weights in this output aggregator can be selected a priori (as is done with batting or slugging averages) or can be estimated statistically based on the performance of the best players in baseball. Once the output aggregator is obtained, an individual player can then be measured relative to the best, and a number between 0 and 1 characterizes his performance as a fraction of the best. The methods are applied to hitters using data from 1995-1999 on all regular players in baseball's major leagues.