axford

ZiPS and the Brewers bullpen — even better?

If you’ve been following the Brewers lately, you’ve probably seen some interesting stuff done with the ZIPS projections, which came out early in the week.  Today, I’m going to try my hand. One of the things I was most interested in going into the offseason was how the Brewers would restock their bullpen, which was among the best in the majors last year and a welcome relief for fans still having nightmares about Derrick Turnbow. Today, I’m going to use ZIPS (and one more system) to look at how this year’s bullpen might stack up with the 2011 squad.

Before we get into the 2012 projections, we first need an objective measure of just how good the 2011 ‘pen was. This isn’t quite as easy as it sounds: Ideally, our measure would be a bit better than ERA, and we also need something that can be replicated with the information that ZIPS gives us (WPA isn’t going to work here). We also have situational importance to account for — John Axford and Kameron Loe had similar innings totals, but due to when Axford was pitching, his performance was much more closely tied with the success or failure of his team.

The first problem can be solved by using a run estimator (I used an average of xFIP and SIERA) in the place of ERA, which we then can multiply by innings pitched to get a kind of revised run total. Next, we multiply the league-average ERA for relievers (3.69) by each pitcher’s innings pitched (in other words, what an average reliever would have allowed in the same innings), which we can then use to figure out runs above or below average. Finally, we can account for situational importance by multiplying this result by the player’s leverage index (John Axford had a leverage index of 1.9, meaning the situations he appeared in added up to being 1.9 times more important than average. Kameron Loe had a leverage index of 1.2.). This gives us a good-enough estimate of how many runs each reliever contributed above or below an average one (which we’ll leave abbreviated as RAA):

IP

RAA

John Axford

73.6

15.53

Kameron Loe

72

10.82

Marco Estrada

51.3

1.62

LaTroy Hawkins

48.3

2.40

Sergio Mitre

33

-1.48

Francisco Rodriguez

29

5.93

Tim Dillard

28.7

1.39

Takashi Saito

26.7

0.63

You might notice that I left out anyone with less than 20 innings in relief — basically, guys who were subject to very small sample sizes and probably wouldn’t have been up if not for injuries. Their innings (84 of the 447 the bullpen threw) need to be accounted for, but how they perform in 15 (generally garbage-time) innings shouldn’t sway our total. We’ll just fill in those 84 innings with replacement level ERA, which gives us a total for the whole bullpen of 32.5 runs above average.

Now, to answer our original question (more accurately, just get ZIPS’ opinion), we need to go through that last process again, only with the numbers ZIPS gives us. Again, there are a few complications: We’ll be using ERA and FIP instead of xFIP and SIERA-they’re the best we can do with the data we’re given, and the projected ERAs don’t carry as much of the random variation that inevitably occurs when the actual results play out. (There are some other relatively minor assumptions we have to make in the absence of 2012 league data, and if you really want to know, leave a comment or email me.)  Here we go:

IP

RAA

Axford

67.3

7.24

Loe

72.3

0.37

Estrada

50

-1.16

K-Rod

64

4.64

Dillard

68.7

-1.02

Veras

63

-1.86

Kintzler

26.7

0.10


The same replacement methodology we used before pegs the projected 2012 bullpen at…5 runs above average. This actually surprised me quite a bit: I wasn’t expecting ZIPS to project them better than last year’s unit, but my guess heading in was that the two ‘pens would be rated pretty close. I can’t get inside ZIPS’ head, but there are a few things to keep in mind before smashing your computer because “it hates the Brewers”:

  • The Brewers had a lot of relievers (Axford, Takashi Saito, LaTroy Hawkins, Francisco Rodriguez to some degree) deliver performances a lot better than their career norms last year, and the ugly truth is that they will probably fall back to earth next year. ZIPS is probably just expecting these guys to fall back to a line more representative of their true talent level.
  • Even with that qualification, the system does seem oddly pessimistic on both Axford and K-Rod. Granted, both are coming off seasons they probably won’t be able to repeat again, and relievers are crazy unpredictable, but both are also yet to have a bad year, and aren’t at the age (they will be 29 and 30 next year) where they would be expected to decline precipitously. Color me confused.
  • The guys (and their leverage values, to some degree) I included in the calculations are there because of how they pitched last year. In this exercise, each pitcher maintains the same workload and role no matter how he actually performs. If Jose Veras or Kameron Loe regress to the point that they aren’t the best seventh-inning option, then a better pitcher will take their innings. We don’t account for this, which probably drags down the total a bit.

Of course, ZIPS isn’t the only projection system out there. I ran the numbers the same way, only using the Bill James projections you can find at Fangraphs, and got this:

(Quick note: Projections for rookies aren’t available, so we’ll have to use Tim Dillard and Brandon Kintzler’s ZIPS lines in our total):

IP

RAA

Axford

72

20.74

Loe

69

0.09

Estrada

51

0.91

K-Rod

67

14.95

Veras

69

0.69

Wow. Thanks to what looks to be incredibly optimistic projections for Axford and K-Rod, the hypothetical 2012 bullpen even vaults past last year’s unit (37.4 runs above average, compared to the 32.5 figure we found earlier).

There’s really not much to be said here: The two systems we looked at have very different opinions as to what to expect next year, and though all we can do is guess, the real result is probably going to be somewhere in between. I have a pretty good idea which system Brewers fans are going to agree with, but you have to remember that both are just useful tools of prediction and not the final word on anything.

(Note: ZIPS projections are from Baseball Think Factory, Bill James projections are from Fangraphs, and leverage scores are from Baseball-Reference. If you have a question, I can be reached at daysofyost@gmail.com.)

Quantcast