Rhode Island College mathematicians Rebecca Sparks and David Abrahamson developed a model which weights different pitcher statistics to predict the players who place first, second and third in Cy Young voting for each league (see November 3 story at http://www.
In addition to yielding the winners, the mathematical model correctly predicted that Dontrelle Willis of the Florida Marlins would come in second place in the National League voting. It also correctly predicted that Minnesota Twins pitcher Johann Santana would be the second-highest-ranked starter in the AL voting results. The model incorrectly predicted that Houston Astros pitcher Roy Oswalt would come in third place (he ended up in fourth place). However, prior to the award announcements, the mathematicians correctly speculated that a better-known player might overshadow the lesser-known Oswalt. And indeed, last year's Cy Young winner, Houston Astros pitcher Roger Clemens, one of the players that the Sparks and Abrahamson had mentioned, pulled ahead of his teammate Oswalt to claim third place.
The model, which is designed to analyze starting pitchers only, is currently not equipped to evaluate relief pitchers, who occasionally win the award. Thus the researchers overrode the model's American League results and had predicted that Yankees relief pitcher Rivera would win the AL pitching title. "We are a little mad at ourselves for not totally trusting the model," Sparks wrote the morning after the AL awards were announced.
The researchers presented their Cy Young model in the April 2005 issue of Math Horizons, published by the Mathematical Association of America. The article is now freely available online at http://www.