Braves Extension Analysis: Craig Kimbrel

This is the fifth in my series on possible extensions for the Braves' young core players. Here are the links to part one (on Martin Prado, alas), part two (Jason Heyward), part three (Freddie Freeman), and part four (Kris Medlen). In the last post, by the way, 73% of TC readers favored extending Medlen at the cost I proposed ($27M over 4 years).

In 2012, Craig Kimbrel posted one of the most mind-bogglingly great relief seasons in history. Few would doubt that he is now the most dominant closer in baseball. Add in the fact that he's only 24 (he turns 25 in May) and has yet to enter arbitration, and Kimbrel is one of the Braves' most valuable players.

Normally, when you have a young player who dominates his position, you'd be eager to lock him up to a long-term extension. However, Kimbrel's position--relief pitcher--adds a huge amount of uncertainty to future projections, which makes such extensions very dicey.

I talked in the last post about how all pitchers are risky long-term, but that goes double (or triple) for relievers, who are infamous for their volatility even in absence of injury. As we'll see below, performance declines among top young relievers are more the rule than the exception.

Still, Kimbrel is a unique talent, and as we've seen the past two years, his peak performance is far greater than most relievers can ever dream of. He could be that rare reliever who's worth the risk of a long-term deal even if he declines from his present levels.

The contract I'm proposing will be a 4-year deal that will buy out Kimbrel's last pre-arbitration year (2013) as well as his 3 arbitration seasons. I'm also tacking on two team option years that would cover his first two seasons of free agency, a fairly common practice that gives the Braves some added upside to help offset the risk of injury or decline.

I'll use the same three criteria from the previous posts to determine a fair value for the contract:

• What is the player's present value?
• What are the chances that he can maintain this value in the future?
• How much did comparable players make in their extensions?

Let's start with the simplest of these questions:

What is Kimbrel's Present Value?

To a large degree, this depends on how much value one places on having a dominant force in high-leverage, late-inning situations. By the common standards of free agent closers, Kimbrel's value would be stratospheric. If Jonathan Papelbon is worth $13 million per season and Rafael Soriano is worth $14 million, then what's Kimbrel worth right now? $20 million per year?

As many more sabermetric-friendly teams and analysts have noted, however, free agent closers and set-up men are overpaid (up to twice as much as any other position in free agency). As great as Papelbon is (by reliever standards), you'd be hard-pressed to find folks outside of the Phillies' front office who think he's going to live up to that contract. It's quite difficult to be worth $10M+ per year if you're only pitching 60 to 70 innings.

Of course, Craig Kimbrel really has been worth $10M+ each of the last two seasons. By Baseball-Reference's version of Wins Above Replacement, he totaled 5.5 WAR in 2011-12; by FanGraphs' WAR, he's been even more incredible, with a total of 6.8 WAR. Even using a conservative estimate of $4.5M per win, Kimbrel has been worth $12-15M per year. Given the even more expensive rates typically paid to free agent relievers, Kimbrel's performance could be "worth" well in excess of $20M per season.

Kimbrel's other numbers are off the charts, too:

* FIP+, like ERA+, adjusts for park and league and sets 100 as average, with each point above 100 equal to 1% better than average. I converted FanGraphs' FIP- stat to FIP+ for better comparison with ERA+.

I think Kimbrel broke those "+" stats. He hasn't just been 50 or 90 percent better than the league average, he's been 169 percent (by ERA) to 213 percent (by FIP) better than average. It's hard to even picture what that means, so let's just say that Kimbrel's pitched about as well as it's possible to pitch.

As you may have noticed, Kimbrel's FIP is even more microscopic than his ERA. While FIP is an imperfect way of measuring relievers (I don't think it's really equipped to handle extreme K rates like Kimbrel's, for instance), it's still much better in small samples than ERA. Certainly FIP gives no indication that Kimbrel's awesome ERA is fluky, or bound to regress toward a less-outstanding mean.

It's a tiny sample--only 160 innings--but what it tells us is that Kimbrel's value isn't so much determined by his performance as it is by the market's upper-limit on reliever salary. Given that the highest reliever salary of all time is the $15M once paid to Mariano Rivera, let's use that number as our baseline. Adding in a 20% extension discount, and Kimbrel's salary in the extension would top out at $12M per year.

Using the 40/60/80 arbitration rule of thumb, a $12M per year value translates to $4.8M in the first arbitration year, $7.2M in the second, and $9.6M in the third. Add in a token raise to $1M for the 2013 season and some option buyouts and we have a total contract value of about $24M over 4 years.

Of course, it's far from certain that Kimbrel will continue to dominate like this for the next 4 years.

What are the chances Kimbrel will maintain his value?

Kimbrel's career has been so singular to date that it is difficult to come up with a good comparison. In the end, I settled on a group of players that had met all of these criteria through their age-24 seasons:

• A total of at least +1 WAR and +1 Win Probability Added (WPA)
• At least 100 total innings pitched and games in relief
• An average of no more than 1.5 IP per relief appearance

The first component winnows the field down to only very successful relievers. The last component ensures that we're not comparing Kimbrel to swing men or '70s/'80s closers who pitched multiple innings regularly.

In the end, I got a group of 21 comparable pitchers* ranging from Francisco Rodriguez to Joe Smith. Here's the full list. Kimbrel's performance is definitely on the high end, but overall it's still a solid comparison group. All of these guys were good at a young age.

* Plus another 6 who were too recent to compare: Joel Zumaya, Chris Perez, Drew Storen, Kenley Jansen, Neftali Feliz, and Aroldis Chapman.

Here are the average stats for the pitchers in the comparison group. WAR/70 and WPA/70 pro-rate those stats to 70 IP, which is about a full season for a closer. (I used B-Ref's WAR and WPA calculations.)

Kimbrel is elite even when compared to this excellent group of players. Only one player in the group (Joakim Soria) had a higher WAR/70 or WPA/70 than Kimbrel (K-Rod had the same WPA/70). No one even came close to Kimbrel's K%, ERA+, or FIP+. All that was bound to happen, though; literally no one can match some of those numbers.

While Kimbrel may well be an incomparable pitcher, it is still useful to see how this group of "comparables" did in the years after this sample. Specifically, we want to know how they fared in their age-25 through age-28 seasons--the years that this proposed extension would cover.

Those are supposed to be the peak years of a typical player's career, yet the comparison group did not improve at all (see the full results for each player here):

The group declined in every facet of a pitcher's game. They walked more, gave up more homers (and hits), and struck out fewer. The end results were still pretty good, but the ERA+ went down by 34 points. Part of that was no doubt a regression to the mean, but the average FIP+ declined by 15 points, so the group actually pitched worse as well.

I should note that the innings pitched average is inflated somewhat by a couple guys--Byung-Hyun Kim and Ryan Madson--who made a significant number of starts. Remove them and the average IP is only 193, or about 48 IP per year on average. In other words, these pitchers put in around 3 full seasons in the 4-year period thanks to injury and ineffectiveness.

That mirrors the results I got for Kris Medlen's comparison group. Apparently if you're giving a pitcher a long-term deal, you should count on him missing about 1 year in every 4. That's just part of the bargain.

Of the 21 pitchers, only Armando Benitez and Joe Smith improved their FIP+ marks (or their ERA+ marks, for that matter). Seven had relatively small declines (<10 points) in their FIP+, and four had modest declines (10-20 points). Finally, more than 1/3 of the group--8 players--had a FIP+ decline of more than 20 points.

The group's performance was similarly dire when measured by WAR/70 (3 improvements) and WPA/70 (4). On average, the group lost 0.8 WAR/70 and 1.2 WPA/70 off their previous levels.

If we price in a moderate decline of 0.8 WAR/70 for Kimbrel relative to his performance to date, he'd still be worth around 1.9 WAR/70 going forward. If we assume he'll be healthy, that's a total of around 7.6 WAR; if we assume he'll miss a year, that's more like 5.7 WAR (or ~1.4 WAR/year on average).

Those numbers imply a free agent salary of just $6-9M per year using the $4.5M/win figure, but taking into account the high price of relievers, perhaps $9-12M per year is more realistic. With a small extension discount, that would imply a total contract in the $16-20M range.

Either way, the Braves shouldn't expect Kimbrel to be as dominant four years from now as he is today, even if he does manage to avoid a major injury.

How much did players like Kimbrel get for their extensions?

Extensions for young, team-controlled relievers are fairly rare. I could only find a few instances in the last 5 years in which a team has extended a reliever with multiple years of team control remaining. The lucky players? Sergio Santos, Carlos Marmol, Joakim Soria, and Manuel Corpas. Here's how their contracts stack up (TO stands for "team option years"):