Baseball Prospectus’ annual PECOTA predictions come out today and have been trending on Twitter all afternoon. This is perhaps the biggest day of the year for Baseball Prospectus, as the predictions will be covered wall-to-wall on blogs and timelines across the web. Since every other site on the internet is getting into the PECOTA predictions themselves, I feel there is need for some context on the subject.
PECOTA, or the Player Empirical Comparison and Optimization Test Algorithm, was a system perfected by stats guru Nate Silver for Baseball Prospectus in 2002. This system has three factors: Major league equivalencies, which take into account minor league stats as a factor in future performance, baseline forecasts that used weighted averages (ERA+, wOBA, etc.) to determine the true skill level of the player, and a career-path adjustment– which uses past players’ stats as an indicator of how similar players’ career numbers can change over time.
This all sounds foolproof in theory, but its is far from it in practice. Last year’s PECOTA predictions predicted just two of the five National League playoff teams correctly– and none in the American League. They were a little more accurate the year before when correctly predicting four of five spots in the National League, but still just predicted one correctly in the American League.
Just because predictions come from computers with sophisticated algorithms, it does not make them any less prone to error than other human predictions. No matter how the models work out, the fact of the matter is is that some things are impossible to predict. Who would have predicted that Daniel Murphy would become Barry Bonds for a week last October?
But in case you were wondering, here are PECOTA’s projected playoff teams:
NL East: Mets
NL Central: Cubs
NL West: Dodgers
NL Wild Cards: Nationals, Giants (It is an even year after all.)
AL East: Rays
AL Central: Indians
AL West: Astros
AL Wild Cards: Red Sox, Blue Jays
The Diamondbacks were projected to finish third in the NL West at 78-84.