October 25, 2010
Checking the Numbers
Lance Berkman had a down year. I know that isn’t exactly earth-shattering news, but he did not perform up to the level we have come to expect given his career numbers. At 34 years old, he is unlikely to continue to hit like he did in the early part of his career, but his 2010 numbers paled in comparison to those produced a year ago, when he hit .274/.399/.509 with a .314 TAv. In 2010, Berkman put up a .288 TAv while hitting .248/.368/.413. Though his season was plagued by injuries, he managed a mere 14 home runs, and that slash line looks strange when attached to his name. The numbers were not terrible, but rather different, considering that he has never posted a BA below .274, an OBP below .386, an SLG below .509, or a TAv below .300.
The 2010 season brought with it plenty of firsts for Berkman, but not in the positive direction. Further, Berkman has averaged around five or six wins above replacement over the last several seasons while amassing just 1.7 WARP this season. The topic of Berkman’s performance came up recently in a discussion of our PECOTA playoff projections. With Mark Teixeira out of commission, Berkman stepped in to play first base and some wondered how the Yankees projected performance was altered with the Astros transplant in the lineup as opposed to Tex. The needle didn’t move much to the point that providing an update to the projection was deemed unnecessary.
This set off some alarm as Berkman had a season worse than any other in his career, and his numbers with the Yankees were especially poor. How could the Yankees be considered as successful with his name penciled in the lineup card instead of Teixeira? Well, this reverts to the topic of a true talent level, which I covered at length a couple of weeks ago. The whole point of using PECOTA to evaluate the various playoff series is that the projection system understands that the most accurate way to develop individual playoff expectations is to look at what a player has done over the last few seasons.
The earlier seasons are weighted less heavily than the more recent ones, but the system grasps that, entering a specific game, the likelihood that Berkman produces certain events should not ignore his past success. This invites the question of whether the Berkman entering the postseason is the guy who surfaced in 2010, or the player who produced from 2007-09. Essentially, in discussing true talent levels, we must reconcile what we feel with what we know. It might feel right to assume Berkman will continue to struggle because he played poorly with the Yankees, but it is much more accurate to expect him to revert to a level of performance somewhere across what he had done over the past four seasons.
But, on the other hand, Berkman’s numbers took a fairly big dip this season. His TAv dropped by 26 points and it seemed as though the injuries had taken their toll on his body and production. My goal here is to conduct a study that could potentially inform a projection system about players whose numbers “crater” relative to a previously established level. Should we expect Berkman to rebound next year to a .310 TAv? That question can be answered, at least on a cursory level, based on the frequency of players who experienced a similar drop in performance from a high level, and how they produced after the year in which their numbers dropped. The answer will not be concrete, but this would be an example of how an extra-curricular study could teach a projection system about what to expect from a certain player. This would be a second level of using comparables.
To that end, I started by pooling together all three-year spans in the Retrosheet era in which the batters stepped to the plate at least 350 times in each season. I then calculated the overall TAv in the span for each hitter. From there, I added the TAv numbers in the fourth season of the span and calculated the difference between that figure and the previously computed three-year average. Using Berkman as an example, his TAv from 2007-09 was .320, while his 2010 mark came in at .288, a difference of 32 points. As much of a dropoff as it seemed Berkman had, he was nowhere near the top of the list. In fact, for players whose average TAv met or exceeded Berkman’s .320, who dropped off by at least 30 points in the fourth season, here are the biggest downturns in performance:
All told, only 47 players fit the Berkman criteria mentioned above, which makes some sense. It wouldn’t be a stretch in the least to assume that incredibly productive players do not commonly decline in a drastic sense. We would expect a more gradual decline, or a general lack of substantially poor seasons. With that pool of 47 players in place, I added the TAv figures for the next season, in order to see if anyone bounced back.
Of the original group of 47 players, three did not return in the fifth season. Of the remaining 44 players, there were 12 that produced worse TAv marks in the fifth year than they did in the year of their performance drop. Another seven players were within 10 points on either side of their TAv in the dropoff year. The other 25 players improved in that fifth season by 12 or more points on their TAvs in the fourth year of the spans.
But what about performance in the fifth year of the span relative to the TAv averaged over the first three years? After all, it is one thing to remain consistent compared to the year of much worse performance, but what is of more interest is whether these players made it back to the previously established and high levels of production. Comparing the TAv in the fifth year to the average of the first three seasons in the span, only five players ended up with a higher mark:
The five players tabled above, much like Berkman, had a very high average, and a down year which was still more than respectable. Still, it doesn’t exactly bode well that 39 of the 44 players were worse than their glory years. What these results suggest is that it is about 50 percent likely that players in Berkman’s shoes will improve in the year following their lesser numbers, but that it is very unlikely that such players will return to the level of performance on display prior to the decline year, at least in the very next season.
Maybe Berkman will latch onto a team this offseason that will give him consistent playing time and put him in a comfortable position, which could help him get back to the Berkman we have come to expect. Or maybe this is the beginning of the end of a very productive career. In any event, it is very interesting to examine the situation, as this doesn’t happen all that often. What do you think? Will Berkman rebound to a TAv in the .305 range? Or should we all get used to the .280-.285 TAv Berkman that could be written onto lineup cards from here on out?