Posts Tagged ‘sabermetrics’


September 30, 2009




With five days left in the 2009 Major League Baseball season, it’s time to look back at some preseason predictions. Who did a better job of forecasting? Was it Las Vegas sports books or baseball sabermetrician/stat geeks?

At least for this season, Advantage Vegas.

Before the season began, the Las Vegas Sun printed the projected win totals of all 30 MLB teams (as forecast by an actual Las Vegas casino.) Vegas got the easy picks right. They saw big years by the Yankees and Red Sox. The casino also correctly predicted that the Angels would win the A.L. West. In the National League, the only accurate prediction by this particular Vegas casino was that Philadelphia would make the playoffs (though they had them in as a Wild Card team.) Picking four of eight playoff teams isn’t exactly a banner season. However, the casino did better than the stat geeks.

For several years, the most accurate sabermetrics system for predicting where teams would finish has been PECOTA. The “Player Empirical Comparison and Optimization Test Algorithm” was devised by Baseball Prospectus. It projects the statistics of every player on a baseball team then crunches those numbers to predict how many wins that team will finish with. Unfortunately, there’s nothing “empirical” about predicting statistics. They are completely made up numbers. PECOTA is simply an attempt to quantify the unquantifiable. It’s a guess (albeit an educated one.)

This year, PECOTA blew it. Take their American League predictions. Please. Sure, they nailed the Red Sox and Yankees, but in the A.L. West and A.L. Central the teams that PECOTA predicted to win the division are actually in LAST place! They forecast playoff berths by the Indians and A’s. That’s about as awful as it gets in terms of predictions. In the National League, PECOTA picked just one playoff team correctly (the Dodgers.)

Of course, if you would have followed OUR preseaon picks, we got as many playoff teams right (4/8) as Vegas.