/cdn.vox-cdn.com/uploads/chorus_image/image/8861059/20120918_kkt_sv7_600.0.jpg)
There is a classic Daffy Duck cartoon called "Duck Amuck" in which a certain wisecracking animator continually redraws the scenery, props and costumes around the hapless water fowl until he reaches the breaking point. This is pretty much what the baseball gods have done to us with respect to Jacoby Ellsbury over the last few years.
Ells broke into the big leagues in 2007 while the Red Sox were cruising toward their second World Series in a four year span, and he hit an enticing .353/.394/.509 in 127 plate appearances. 2008 was his first full season and while he didn’t repeat those crazy ’07 numbers his .280/.336/.394 line was very good for a first year regular; he looked excellent in both left and center on defense as well. fWAR had him at 4.2 wins for that season, star-caliber production. His bat took another small step forward in 2009, but that cruel prankster with the pencil erased his glove and replaced it with an anvil, and his poor defensive metrics dragged his fWAR down to 2.4. In 2010, the center field background was swapped for the Green Monster of left just long enough for Adrian Beltre to give Ellsbury's rib cage the accordion treatment. He was never fully healthy again that year and he saw just 84 plate appearances.
Ells returned with a vengeance in 2011, destroying American League pitching to the tune of .321/.376/.552 with 32 home runs and 39 stoles bases, finishing second in the MVP vote. However, just when it looked like he would soar to free agency and a Matt Kemp-sized deal, the plane disappeared and his parachute was replaced by a Reid Brignac-shaped boulder. He missed almost the entire first half of the 2012 season and hit .271/.313/.370 in 323 plate appearances for the year.
This mix of greatness and injury-induced mediocrity is a problem when it comes to projecting Jacoby Ellsbury. These days there are many great projection systems to choose from and each one handles things just a little bit differently, but they all have one important trait in common; they are all generated entirely from the statistical record the game has left us. They vary in the weight that they give to certain numbers, but they are systematic and they do not make special adjustments for individual cases. That is their beauty, after all. Players have a baseline of talent and their performances vary around that baseline. For Jacoby Ellsbury, that baseline is extremely clouded by all that lost playing time. To illustrate this point, I made up three simplistic Marcel-style projections.
Season
|
PA
|
AVG
|
OBP
|
|
wOBA
|
wRC+
|
HR
|
SB
|
fWAR
|
Marcel-style
|
412
|
0.272
|
0.320
|
0.405
|
0.318
|
94
|
13
|
21
|
3.8
|
Healthy Marcel
|
694
|
0.306
|
0.361
|
0.475
|
0.364
|
120
|
19
|
52
|
6.0
|
Injured Marcel
|
258
|
0.249
|
0.293
|
0.336
|
0.279
|
67
|
3
|
12
|
1.0
|
The first line is the average of the past three seasons with a 3-5-7 weight applied. Apart from the stolen bases, these numbers are not far off from what Jacoby Ellsbury did in 2008 and 2009. 2008 was Ellsbury’s rookie season, however. He was just 24, with fewer than 200 at bats and he was already that good. Typically, we expect some progress between a player’s rookie season and his peak years. That brings us to the "Healthy Marcel line. Here the same weights are used, but only the last three seasons where Ellsbury was healthy are considered. This is the player we might have expected Ellsbury to become if 2010 and 2012 had never happened. Here, he comes close to 700 plate appearances, improves his on-base skills and his power and stays near the top of the stolen base leader boards. Sadly though, there is the dark side to consider; Following injuries, he has not been the same player at all. We can’t just ignore these numbers, but it also doesn’t really make sense to give them the same weight as his injury-free performances.
These numbers are just to highlight the contrast and despite the Marcel-like process, there is no reason to think of them as real projections when we have plenty of real projection systems to consider. Fangraphs now publishes four such systems and a crowd-sourced set and they have an interesting range of numbers for the Red Sox center fielder-
Projection
|
PA
|
AVG
|
OBP
|
|
wOBA
|
HR
|
SB
|
WAR
|
Career Avg
|
467
|
0.297
|
0.349
|
0.442
|
0.346
|
10
|
34
|
3.4
|
Steamer
|
597
|
0.285
|
0.340
|
0.431
|
0.335
|
14
|
22
|
4.3
|
Bill James
|
668
|
0.294
|
0.346
|
0.436
|
0.340
|
15
|
37
|
4.9*
|
Oliver
|
460
|
0.285
|
0.343
|
0.441
|
0.340
|
11
|
18
|
3
|
Fans (44)
|
631
|
0.295
|
0.349
|
0.440
|
0.344
|
16
|
34
|
5.1
|
ZiPS
|
451
|
0.284
|
0.335
|
0.445
|
0.337
|
12
|
23
|
2.9
|
Average
|
546
|
0.290
|
0.344
|
0.439
|
0.340
|
13
|
28
|
3.7
|
*The Bill James projections do not include defensive numbers so the fWAR here is estimated.
I have highlighted the highest and lowest numbers. The fans are the most optimistic about Ellsbury, followed closely by the Bill James system which is very similar to what fans think except in the slugging. Oliver and Zips are both low on Ells, both in terms of totals and in rates. Steamer projects more playing time than those but has essentially the same take on his rate stats. Every system projects Ellsbury to be an above average defender, but the impact varies. The consensus is that Ellsbury will be an above average player, but only James and the fans see him approaching star-level next year.
These are all excellent projection systems and with numbers so close to his career norms, it is hard to take issue with the process, but it is also hard to believe that a healthy Jacoby Ellsbury would not beat even some of the more optimistic numbers above. Playing time is the most difficult to thing to estimate, so it is a bit unfair to get too caught up in it, but it is hard to see how Ellsbury would not approach 700 plate appearances. His career has been full of extremes so he has never been within 100 plate appearances of the average of these projections. When healthy he bats leadoff and averages 694 plate appearances. He had just 323 last season, but that came in half a season. If he doesn’t match at least James’ projections, he probably won’t top 300 plate appearances either.
No projection system has Ellsbury topping or even matching his career batting average. That is understandable, since averages tend to decrease as players age and Ells’ K% is creeping up, but he is still at his peak and with a .322 career batting average on balls in play, an over bet here would not be outlandish. No on-base percentage projection has him topping his career average either and that seems more problematic. Ellsbury has had three full, healthy seasons in his career and his on-base percentage has increased in each one. As a rookie he was close to the average of these projections. Clearly, his 2012 performance is dragging down these numbers. If that season was not given so much weight, at least a few people would project rates above his career averages. Systematically, it makes sense to take that sample as a sign of decline. In this case, I think that is being overstated.
Power and stolen bases are the ultimate question marks for Ellsbury. His 32 home runs in 2011 were a surprise and it is certainly not a good idea to project that type of power again. It is not surprising then that everyone is projecting slugging numbers very close to his career levels. Conversely, Ellsbury had been a top base thief but his age and a recent history are both pointing to fewer steals in 2013. The projection systems reflect that, but the playing time issue exaggerates the effect. In our "healthy marcel" we had 52 stolen bases for Ellsbury, but that is would be a surprisingly high number at this point. Both numbers are almost pretty well unpredictable at this point.
2013 will be an extremely important season for Ellsbury, since he will be a free agent following it. If he does manage to play close to 162 games, it also will be very interesting to see if he is the player we might project had those injuries not happened or if he now starting to level out or even gradually decline as the more negative projection systems seem to suggest.