Where Sabermetrics Fail

The Baseball Prospectus has issued its annual PECOTA projection (seen here) and, as usual, it is highly controversial. PECOTA (Player Empirical Comparison and Optimization Test Algorithm) is a sabermetric process for projecting the performance of individual players and teams over the course of the upcoming baseball season. In the past, these projections have been quite accurate, generally, for individual players, but often laughably off for teams. Last year’s projections for the American League proved almost completely wrong, whereas the National League projections were almost exactly right. This year’s NL projection looks quite reasonable, but once again, the AL projection seems way off. For instance, two of last year’s powerhouses, the Texas Rangers and the World Series Champion Kansas City Royals, both of which made few changes in the offseason, are projected to finish at or near the bottom of their respective divisions.

The Flaw in the Methodology

I believe the flaw in the methodology is that sabermetrics are designed to evaluate individual players, and do not take into account the combination effects of players on a team. In particular, the method tends to devalue pitching, because it sees the function as run suppression, but ignores the emotional impact that team pitching has on the rest of the team. When pitchers are doing well, holding back the opposition, their teammates feel more confident at the plate, getting more base hits, and more extra base hits. In the field also, position players feel more comfortable taking chances to make great plays instead of taking the safer route. By contrast, when the pitching is poor, hitters become anxious, not wanting to waste an at-bat, often going deep into counts only to strike out or pop up. Fielders are tense, afraid of screwing up, and may not dive after that ground ball or leap for a line drive. The sabermetric guys will look at the season-ending stats and say, “well of course they lost… they failed to make plays, and their OBP was down.” But there is a cause-and-effect relationship between good pitching and good play by the rest of the team.

How Pitching Problems Can Make a Team Unravel

I myself saw a very visible example of this in the 2011 Red Sox. The pitching was shaky during spring training, and they had a terrible April. But despite a terrible performance by John Lackey (who, apparently, was hiding an injury), the pitching staff got it together behind inspiring work by Clay Buchholz, and the All Star break saw the Sox with a seemingly insurmountable lead in the AL East. Other injuries began to take their toll among the pitchers after that. Tim Wakefield, who was near the end of his career and supposed to be in the bullpen and provide an occasional spot start, found himself in the regular rotation. By the end of July, he was visibly worn, and his flat knuckleball was routinely being launched into the cheap seats. When Clay Buchholz suffered his season ending injury, it was the last straw. You could just see the energy flow out of the team. They looked unsure of themselves in the field, and lackluster at the plate. It was plane that the team was nosediving out of the season altogether. Some people blamed manager Terry Francona, and some blamed the beer and chicken incident. For me, the beer and chicken incident was an effect, not the cause. The cause was the collapse of effective pitching.

Last year, many people picked the Sox to win the AL East. I was one of them. We all put too much stock in the lineup. In fact, there was no ace in the pitching staff. Rick Porcello was a decent, workmanlike low ball pitcher, but he was no ace, and putting him at the top of the rotation put more pressure on him than he could handle. Clay Buchholz had his usual midseason exit, and the pitching sucked, overall. The team looked sluggish and uninspired, and dramatically underperformed as a result.


So, why is the PECOTA projection for the NL so good, and for the AL so bad? Simple. The NL is a hitters’ league, so the teams are less affected by the performance of the pitching staff. The AL, on the other hand, is a pitchers’ league, so the effect I mention is very strong there.

The analysts who do these projections need to do a better job of taking the interplay among teammates, and the emotional component of the game, into account when making projections. Ballplayers are not like robots in an assembly plant. When things are going well, they do well. When things are going badly, they slump. They are, in fact, human beings.


Leave a comment

Filed under Boston Red Sox, Major League Baseball, Rangers, Red Sox, Royals, Texas Rangers, Uncategorized

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s