A Mathematical Model to Predict Award Winners

Baseball has always been a delight for the mathematically inclined. Numerical records have been integral to following the sport, and popular measurements like batting averages have enticed many youngsters to experiment with simple calculations. In the last three decades, a great many fans and even some baseball executives have brought significant mathematical sophistication and the scientific method to the study of baseball. Evidence of this movement appears in publications ranging from the daily sports page to this magazine. We know that a player’s offensive contribution can be measured by things such as his slugging and on-base averages; we know that a team’s won-lost record can be predicted from its totals of runs scored and runs allowed. There have also been many discussions on using mathematics to compare players across eras. Organizations like the Society for American Baseball Research (SABR) have welcomed mathematical analyses; indeed, the name SABR has given rise to the term “sabermetrics” to describe the mathematical study of the various facets of the game. In this article, we wish to follow a somewhat uncommon direction. Rather than analyzing how some measures of onfield performance correlate with other occurrences within the game, we want to see how they predict an off-field assessment. When a ballot is cast for an award—the Rookie of the Year, the Most Valuable Player, and other such awards—the voter presumably has some criteria in mind. One voter may pay particular attention to batting average, one may emphasize runs batted in, one may look primarily at where a player’s team finished in the standings, and so on. A voter may or may not publish his criteria; indeed, he may very well be unaware of exactly what those factors are as he forms his impressions of various players. When the ballots are tallied, the voters will have ranked the candidates for an award, from first place on down. Every year there are an untold number of spirited discussions of how voting will go, with predictions from all corners about who will win. One cannot help but wonder if a voting result is in fact predictable from the data available to the voters. Is it possible to look at the information for each candidate and combine it in such a way that we can predict how the voters will compare him to the others? In at least one case, we believe the answer is in the affirmative. In what follows, we will present a way to predict the ranking of candidates for one major award.