Change-UpJanuary 14, 2011
Bobby Jenks & Kyle Farnsworth
By Patrick Sullivan

Well, this is post number two this week provoked by Dave Cameron's writing. The funny thing is that I really liked both pieces. I touched on a subtle point about tone with regard to defense and Wins Above Replacement on Wednesday. This morning, I want to address Dave's latest. As I mentioned, it's a good piece. Dave compares Kyle Farnsworth and Bobby Jenks, their respective last two campaigns, the contracts they signed, and the general reaction to both. I am Exhibit A for Dave. I loved the Jenks deal, I panned Tampa Bay's Farnsworth signing. I should also mention I am a Red Sox fan.

Here's Dave:

News Item #1 – AL East team with strong sabermetric leanings signs free-agent reliever. His peripherals are better than his ERA, and he’s considered to have some personality baggage. Reliever gets $12 million over two years – reaction is mostly positive.

News Item #2 – AL East team with strong sabermetric leanings signs free-agent reliever. His peripherals are better than his ERA, and he’s considered to have some pesonality baggage. Reliever gets $3.5 million for one year – reaction is abject mocking.

As you may have figured out by now, News Item #1 refers to the Boston Red Sox’ signing of Bobby Jenks, which took place a month ago, while News Item #2 refers to the Tampa Bay Rays’ signing of Kyle Farnsworth, which broke today. I find the differences in response to these deals somewhat amusing.

Here are the numbers for Farnsworth and Jenks over the last two seasons.

Farns: 102 IP, 2.91 BB/9, 9.09 K/9, 0.62 HR/9, 43.0% GB%, 3.79 ERA, 3.08 FIP, 3.54 xFIP
Jenks: 106 IP, 2.89 BB/9, 9.34 K/9, 1.02 HR/9, 53.3% GB%, 4.08 ERA, 3.53 FIP, 3.13 xFIP

Go read the rest of it. He touches on changes in Farnsworth's delivery, and it's a nicely written and evidenced post about how there may not be much difference at all between Jenks and Farnsworth. Here is how he concludes it, however.

Once you factor in the size and length of their respective contracts, it seems pretty clear to me that the Rays got a better deal with Farnsworth than the Red Sox did with Jenks.

I find this conclusion problematic for a few different reasons. The first has to do with the innings Jenks and Farnsworth have been pitching. Let's first look at Leverage numbers, a metric tracked at Baseball Prospectus. 1.00 is the Leverage situation at the start of the game when the first pitch is thrown, and then from there it's driven by Win Expectancy. We will limit our look to Dave's comparison of 2009 and 2010 for obvious reasons. Going back further overwhelmingly favors Jenks, and it does seem that Farnsworth may have turned a corner with regard to mechanics. So 2009 and 2010 only it is.

Player 2009 2010
Jenks 1.65 1.43
Farnsworth 0.86 1.20 (KC), 0.94 (ATL)

As you can see, Jenks has been pitching very important innings over the last few years. Farnsworth, not so much outside of a handful of key appearances for the Royals in 2010. It's important to remember, too, that Jenks has been closing games for a perennial contender while Farnsworth has pitched for the Royals. His ~20 innings with Atlanta in a pennant race were often low to medium leverage situations. If there were a playoff expectancy or championship expectancy figure, the gulf would be even wider.

Farnsworth's high leverage innings have come amid a push for 70 wins, while Jenks's have come in a pennant race. But let's set that aside for the moment and just look at how they have performed in their respective high leverage situations. Here it is, presented as OPS against.

Player 2009 2010 Career
Jenks .677 .813 .660
Farnsworth 1.422 .802 .736

Let's remember that Dave's conclusion on its face makes the narrow point that $3.5 million guaranteed for one season to Farnsworth is better than $12 million guaranteed to Jenks over two seasons. It's not terribly provocative in the context of how he presents it. They have similar peripherals, and there's reason to believe Farnsworth's improvement in 2009 and 2010 is real. Over and above the leverage point I have made above - both the innings pitched in those situations and how they have performed once there - there are additional considerations.

A $3.5 million investment for the Rays accounts for a greater percentage of their payroll than a $6 million annual investment does for the Red Sox. It's great that Tampa Bay has carved out a niche developing talent and finding undervalued assets but, just like the Boston Red Sox, the Rays are in the winning business. I understand they do not have the luxury of ponying up $12 million for someone like Jenks. In that context, given their similar output in 2009 and 2010, Farnsworth makes for a nice proxy. But let's not jump to the conclusion that the market is out of whack, or that Boston missed on Farnsworth by paying up for Jenks. Jenks comes at a premium for a number of reasons, and it's not just because professional evaluators only remember Farnsworth's failures as a Yankee.

Jenks is 30, Farnsworth 35. Jenks has five career postseason Saves, including two in the 2005 World Series as a rookie. Jenks has a quality track record extending back to 2005, while Farnsworth optimism hinges on 102 largely meaningless innings in 2009 and 2010. I don't like to overemphasize "clutch" statistics but when it comes to evaluating relief pitching I think all the information we can get is relevant. It's particularly so in the ultra-competitive American League East. In this light, when you consider age, leverage and career quality, even if the Rays have unearthed another gem in Farnsworth, the respective contracts look about right to me.


Leverage measures how the manager chose to use pitcher. Jenks was used in important innings, Farnsworth less so. I don't think I would give that much weight in projecting what they are likely to do going forward. More important is the age question, and the question of whether or not Farnsworth has made some real adjustments.

There is little (if any) evidence to suggest that it is more difficult to perform in a high leverage situation than a low leverage situation.

The fact that Jenks was being used in higher leverage situations doesn't really help predict his future performance in any situation nearly as much as using his actual performance. Similarly for Farnsworth, it isn't his fault he was on a bad team. He's pitching to the same hitters Jenks was and getting the same results.

BPro LI is notoriously bad.

First comment touches on my point: both of these guys are going to be 7th/8th inning setup men. Cameron proves their track record is similar in terms of output, and heck, even pitch types and styles. Is this just a way to get Jenks in the door to compete with Bard in 2012 for the closer role? Otherwise, this is a siginificant overpay for Jenks.

One quick point, which Sully and I quickly talked about on Twitter yesterday; the argument for clutch as a skill worth paying for is a very slippery slope.

For instance, let's take Farnsworth's Clutch score (as calculated by FanGraphs). Patrick's right that he's been absolutely awful in high leverage situations the last two years, but he was actually better in high leverage situations the four years prior, and his career Clutch rating is just a tick below average (due entirely to the last two years).

The reason I limited the statistical comparison to 2009 and 2010 was because that's when Farnsworth altered his pitching repertoire and added the cutter, so there is some evidence that he's become a somewhat better pitcher than his career numbers suggest. However, I don't see the reasoning for limiting the Clutch sampling to just those same two years.

There's no reason to believe that adding a cutter makes you worse in high leverage situations, at least none that I can tell. If we're going to argue for Clutch as a skill, then we need as large of a sample as possible. Reducing our window to just 100 innings renders the numbers essentially meaningless, since we're already hunting for something that is shown to generally not have any predictive value.

I don't think leverage has any direct effect on their respective performances; However:

1. Farnsworth has been homerun-lucky over that time- his HR/9 doesn't squre with a flyball pitcher.

2. Jenks has been extremly unlucky in homeruns, behind only Brandon League in HR/FB over that span.

3. Farnsworth has been platooned to some degree, facing mostly righties; while Jenks has not.

"There is little (if any) evidence to suggest that it is more difficult to perform in a high leverage situation than a low leverage situation."

Historically and over the entire game of baseball, that's true. But some individual pitchers struggle in hi-lev situations, while others thrive in them and struggle when the leverage is lower. I don't think it's a stretch to say that Kyle Farnsworth is a pitcher who statistics and observation agree fails to perform well in high-leverage situations.

I still think Cameron's biggest mistake is in dismissing the earlier performances of each pitcher and assuming that both will pitch, going forward, in a way that reflects the past two years.

I would stake my life savings that that won't be the case.

One common theme on sabermetric websites is that pressure has no bearing on performance (i.e. clutch performers). However, GMs and managers seem to value those who have demonstrated clutch performances (yes for Jenks versus no for Farnsworth) and pay and play players accordingly (more $ for Jenks vs less $ for Farnsworth).

This is one of the areas that sabermetricians and "old school" writers derisively attack each other. However, this probably means that the truth lies somewhere in-between.

Mark, re #1, Farnsworth has been a groundball pitcher since he added the sinker and changeup and increased his cutter usage dramatically starting in 2009.

PaulSF, sabermetricians who are skeptical of whether we can identify players with more or less clutch skill do not disagree that there are clutch performances. The question is whether clutch performance in the past is any predictor of clutch performance in the future.

Also, you mention statistics and observation, but you don't mention what observation you're talking about, unless it's simply observing the same performance that the statistics already reflect.

Mike, Farnsworth has a 43% groundball rate since 2009 and only a 41% last year. both numbers solidly below average.

Mark, I probably should have said he became "more of a groundball pitcher". He went from 38% groundballs in his prior career to 44% in 2009-2010 (per Retrosheet/MLBAM). League average is 44%, so he's right at average now.

Since he changed his repertoire so much from 2008 to 2009, I think it's fair to investigate his performance since then and see if the reduction in HR/9 is due to his new approach. I haven't seen anyone do that yet, but I certainly wouldn't reject out of hand the idea that the increase in GB% caused the drop in HR/9. We'd also need to look at whether his new pitches affected the ability of batters to pull the ball off of him.

"But some individual pitchers struggle in hi-lev situations, while others thrive in them and struggle when the leverage is lower. I don't think it's a stretch to say that Kyle Farnsworth is a pitcher who statistics and observation agree fails to perform well in high-leverage situations."

If every pitcher was exactly identical, when you looked back in the past you'd expect to see a bell shaped curve if graphing "clutch" performance with some players much better and some much worse than the mean. Going into the future, however, you know they are all identical. The ones that were more clutch aren't any more likely to do better in the future than the ones that were less clutch in the past.

So yes, the smaller the sample size you look at for a given player (i.e. relief pitchers for 1 or 2 seasons here or there) the more likely you are to find a deviation from the mean that is the result of randomness and not talent.

Not really the point, but I only remember mocking Farnsworth to the tune of Mariano Rivera's chosen entrance music:

Enter Kyle
Good bye smile!
His pitch lands
Far beyond the grandstands

Jenks would be the clear choice, regardless of the leverage analysis, if he dedicates himself to getting back in shape. The move to Boston should provide the motivation needed. They are loaded for a title run in 2011. A setup role will take some pressure off Jenks.