Forget traditional scouting reports; in order to predict the 2011 MLB season, one New Jersey Institute of Technology professor says all you need to know is math.
Bruce Bukiet, associate professor and associate dean of mathematical sciences at the New Jersey Institute of Technology, uses applied mathematics to determine which MLB teams will have the best year. After a wave of high-profile free agency moves including acquiring sluggers Carl Crawford and Adrian Gonzalez, the Boston Red Sox are Bukiet's team of choice along with their archrivals, the New York Yankees, as both, he says will win 97 games apiece.
Along with the Red Sox and Yankees, Bukiet has the Tigers and Rangers each winning their divisions to earn a playoff spot. In the NL, he has the Phillies, Braves, Cardinals and last year's World Series champ, the Giants, in the playoffs.
Bukiet uses a linear algebra algorithm which works off the best batting order for a team and the likelihood each player will single, double, triple, hit a home run, strike out or make an infield out during an at-bat. For each batter, he computes the likelihood that batter will hit in certain situations. Using those numbers he can calculates how many runs the team will score. Then for entire teams he computes how their lineup would fare against others.
Once he factors in pitchers and the team's bench, he figures out how often one team will win against others, and factors in home-field advantage. At the end of his series of calculations he gets the team's total win count.
For each team, to run through the whole season takes me about five minutes, Bukiet said. My son, who started doing this with me when he was 15 and is now 26, has done a lot to help me automate it. He's written it on the internet, put in a file of each player's data, who is there and who is not there, etc. For the first few years, I was doing it by hand. Now it's automated.
While the idea to use math to predict the baseball season has been his head since 1988, Bukiet has been applying his methodology for 10 years and says for the most part it has been effective. Out of the 10 years, his method has been better than those of the oddsmakers seven of them. The other three years he was slightly off. Bukiet has published an enhanced model of the method in the International Journal of Performance Analysis in Sports.
Over the years, he has tinkered with the methodology, adding other factors along the way. For pitchers, he calculates their effectiveness through two-thirds of a game. He then applies the team's bullpen average for the rest of a game. Unlike other sports, Bukiet says baseball works very well.
There's a lot of data associated with the sport and the one thing that makes baseball doable is there are a lot of one-on-one battles. In basketball, there are five guys. In football, there are 11 guys on each side, Bukiet said.
At the end of each season, Bukiet describes the probability for each postseason series. He doesn't do this in the beginning of the season because of the likelihood there will be trades. Or as Bukiet, a longtime Mets fan, said, Teams like the Yankees will just buy people.
Unlike the regular season, he says his method is not as effective for the postseason. This is because the series are shorter and random factors play a larger role. All of his results, including how he has fared in the past, are published at http://www.egrandslam.com.
Over the past few years, numbers have become an increasingly important part of baseball. A 2003 book written by Michael Lewis called Moneyball detailed how Oakland Athletics general manager Billy Beane used number crunching to build his team. In addition, every year, MIT hosts the Sloan Sports Analytics Conference, on the role statistics play in helping sports teams build a succesful franchise.