One of the weird things about being a sports fan or a sports writer is that there are moments when it fills you with a sense of entirely unjustified pride. Just such a moment arrived for me on Tuesday when the White Sox signed left-handed reliever Zach Duke to a three year, $15 million deal. I don't know Zach Duke personally and I honestly probably wouldn't recognize him out of uniform, but I developed a little bit of a Zach Duke obsession this summer because Duke was doing this crazy thing where he was morphing into an entirely different pitcher before our very eyes.
Now you're probably wondering why Duke gets as many words as the Jason Heyward swap or the Russell Martin deal. This is a middle reliever signing with a club that's probably not quite a wild card contender in 2015. It's the kind of deal that gets attention because it's November but not the kind of deal that requires real analysis.
Except Zach Duke had my attention all summer because his strikeout rate erupted this season. He worked hard on his mechanics last winter, added a lower arm slot, and started firing off more breaking balls (sliders and curves) than he had ever thrown before. It was one of those big changes in performance that you could tie to a reasonably clear change in process. I didn't really expect the strikeout rate to stay that high, but I bought into the overall trajectory.
Zach Duke had reinvented himself! I actively sought out updates on Zach Duke appearances. It wasn't the Corey Kluber Society or anything, but I felt some degree of ownership.
Entering 2014, he had a career high strikeout rate of 17.9% that came in a tiny sample of innings in 2012. Last year, he had one month with a strikeout rate below 29%. Before we get ahead of ourselves, here is Duke before 2014 and during 2014:
It's obviously only one season's worth of innings from a reliever, but it's about 60 fantastic innings not heavily distorted by a platoon. He faced a roughly equal number of righties and lefties and had pretty similar results. I'm not going to sit here and argue that the thousand innings that came before should get thrown out or that he's a true talent relief ace based on one year of data, but he's demonstrated the ability to perform at a high level with no BABIP luck, HR/FB% magic, or really well targeted appearances against weak lefties. He legitimately pitched well for those 60 innings.
It doesn't mean he's going to do it again, though. It just means he's capable of it. Steamer sees a big drop in strikeout rate (to about 21%), but that's still nearly twice as high as it was in his entire career before 2014. In other words, Steamer is buying that he's much better. I'm suggesting that he's much better. No one is sure about the magnitude of that increase because this isn't exactly the most common career path.
A multi-year deal for a 31 year old lefty who was a washed up starter in the not too distant past seems kind of crazy on its face, but to provide $15 million of value over three years he only needs to get to something like 2-3 WAR. If you're a 0.7 WAR reliever every year you're set.
Granted, that's not exactly a sure thing for a guy who hadn't been worth more than 0.7 WAR since 2009 (when he was a starter). It's a big commitment that's banking on the new Duke. He doesn't have to match last season, but he has to be good for this to make any sense. We talk about how relievers are unpredictable quite often, so I thought a simple test would be in order.
This is a three year deal. Duke was an awesome pitcher this year. How well do awesome relievers perform over the next three years?
I grabbed data from 2011 and then 2012-2014 for all pitchers with 30 or more innings in relief in 2011. That gave me a sample of 200 relievers who I divided up into four "equal" groups based on FIP-. Here are the FIP- of the groups (weighted by batters faced) in 2011 and then in 2012-2014.
|Group (FIP-)||Number||2011 FIP-||2012-2014 FIP-|
|Tier 1 (0-80)||49||65||83|
|Tier 2 (81-92)||50||87||94|
|Tier 3 (93-109)||50||101||100|
|Tier 4 (110+)||51||122||102|
Two things about this chart seem important. The first is that there's an obvious survivor problem in Tier 4. The pitchers who are really terrible get far fewer chances in the next three years than the ones who are decent. Second, the top group takes the biggest step back, which seems like a win for our old friend regression toward the mean. But it's also worth noting that they were still a very good group, even if they weren't elite and performed worse.
If the Sox get three years of an 85 FIP- from Duke they are going to be over the moon. It's also a pretty clean distribution. Of the 49 pitchers in Tier 1, 23 were at 86 FIP- or better over the next three years and 24 were worse (two didn't pitch).
Could Duke turn into Kameron Loe (129 FIP- from 2012-2014)? Sure. But he could also wind up being Joaquin Benoit (81 FIP- from 2012-2014).
There are certainly complicating factors like age and track record and usage, but a quick test suggests that judging relievers on one year of data isn't totally pointless. The best relievers typically perform better over the next three years than the rest of the league, even if they don't maintain their dominance in quite the same way.
Zach Duke's going to be an interesting case. He seems like someone who should do well going forward and history suggests he won't become totally worthless. But 365 days ago, not one team would have even considered a deal like this. The Brewers got him for nothing, he turned into something, and now he's a very rich man.
We'll see if he can keep it up.
Statistics courtesy of FanGraphs.