Guest Author - Joe Mancini
The last time we met we started a discussion of how clubs (and prognosticators in the media and Las Vegas e.g.) attempt to “forecast” the performance of teams and players over an upcoming season using a variety of ever-increasingly complex statistical and mathematical models.
PECOTA (Player Empirical Comparison and Optimization Test Algorithm) is one of the more widely-known forecasting methods; it grew out of the trailblazing work of Bill James and the other “Sabermetricians” of the Society for American Baseball Research. Most big-league teams (if not all) employ their own so-called “stat-geeks” and develop their own, proprietary forecasting models, not unlike, say, econometricians at major banks and financial and investment firms. Fantasy and rotisserie players frequently employ the PECOTA predictions in determining which players to draft; and the pre-season plethora of magazines about the upcoming season are increasingly aimed at fantasy players.
I don’t know if anyone regularly evaluates such projections against actual results, the last I saw 2007’s PECOTA projections were not bad with regard to hitters (over 60% correlation), not as good for pitchers (somewhere a little over 40%). What it really gets back to is the old saying, “It’s baseball and you never know.” Projections, predictions, all that is fun and all, but none of it should be taken too seriously. Baseball has two salient features: it is the most difficult of all sports in which to project performance, and that performance almost always “regresses to the mean”. Yes, that’s another statistical formulation which states, basically, that things tend to average out, in other words, Albert Pujols isn’t going to go 0-fer forever, and the Washington Nationals’ winning streak, say, isn’t likely to go past 10 games. Those statements are true in just about every case.
And then, of course, the unexpected happens, Adam Wainwright goes down for the Cardinals, Chase Utley can’t answer the bell for the Phillies, and now outlooks for those clubs become considerably cloudier. Zack Greinke breaks a rib playing pickup basketball and the Brewers, who couldn’t wait for Opening Day, are now filled with foreboding. It’s baseball, and you never know.
In his best-seller, “The Black Swan: The Impact of the Highly Improbable”, Nassim Nicholas Taleb documents at great length how despite the best efforts of people to foresee, model, predict, project, what really happens is almost always stunningly different. We can’t even properly understand today’s events in historical context. He refers to it as “The Triplet of Opacity”, to wit, 1) the illusion of understand (we think we understand what’s going on), 2) retrospective distortion (past events seem to make more sense after the fact than during), and 3) the overvaluation of factual information, especially when we try to “Platonify” or categorize. Certainly all those factors come into play when we’re trying to get a grip on events that are soon to overtake us.
I promise, next week we’ll be chattier.