The Regression of a Bullpen

One of the great stories of the 2008 season for the Rays was their much improved bullpen following a historically dreadful 2007. The renaissance was largely led by the emergence of JP Howell and Grant Balfour. In October of 2008, Peter Bendix of Beyond the Boxscore wrote what appears to be a pretty prophetic piece entitled "Assessing the 2009 Tampa Bay Rays Bullpen" where he called for a serious regression.

Bendix pointed out that the team's BABIP should somewhat resemble .290 based on the team's DER of .710. Below is the data put together in 2008 by Bendix compared to 2009's numbers using the metrics of BABIP, HR/FB, and LOB%.  I took the liberty of using Brian Shouse for Trever Miller, Lance Cormier for Jason Hammel, and Joe Nelson for Chad Bradford as these new pitchers assumed their predecessors' roles.  

Interestingly the three newcomers all have lower BABIPs than their predecessors. All four returning pitchers have increased their BABIP significantly by an average of .129.  That is a huge margin. A portion of that can probably be attributed to playing half our games against the Red Sox and Yankees to date. The Red Sox have a team BABIP of .325. These figures should regress handsomely.

BABIP covers when the ball is play. Let's take a look at the fielding independent stats HR/9, BB/9 and K/9 to see if this is merely BABIP regression or if maybe there are bigger concerns.

The bullpen is giving up homeruns at a much higher clip while strikeouts and walks are fairly flat. If we dig deeper and remove the new guys from the equation, we see that the returning bullpen arms are struggling beyond a BABIP regression.

 Bendix argued that the relievers'' ERAs were too low relative to their FIP in 2008. That's not good news when the FIPs are up as well this year.  On the bright side, it does seem that we have made significant strides in upgrading from Bradford, Miller, and Hammel to Nelson, Shouse, and Cormier.

Last season our bullpen inherited runners at a clip of .47/IP. Of those runners, 25% scored.  In 2009, the pen is inheriting runners at a clip of .70/IP. Those runners are scoring 31% of the time.  In summation, we are inheriting .23 more runners per inning and allowing 6% more of them to score. Not a good mix.

Last year the bullpen surrendered .386 runs per inning. In 2009 they are allowing .46 runs per inning.  If we remove the newcomers and look just at the returning 4, their runs allowed/inning jumps from .35 in 2008 to .57 in 2009.

It's very interesting to note that Grant Balfour has allowed 9 runs to score in 11.33 innings pitched. However when he is brought in mid-inning with inherited runners he has allowed none of those runners to score. That is why he is the firefighter.  When he begins an inning, he gets bored and starts a fire. I've always felt he's best used to work out of a jam or as a closer if no jam occurs.

In summation, our bullpen struggles can be attributed to an expected BABIP regression from 2008, a difficult schedule to begin the season and fielding independent pitching struggles beyond 2008 levels by the returning four. On the bright side, the schedule gets easier, the BABIP correction has been extreme and the new guys have mostly done fantastic. I wonder if some of the FIP struggles can be attributed to familiarity of opponent. Most of Wheeler's struggles have occurred against the Yankees, Howell vs. the Red Sox and Yankees, and Balfour vs. the Orioles and Red Sox, all divisional opponents.  Have there been past studies done to compare the success of bullpen arms changing division or becoming stale? Bullpen arms are usually pitchers who you don't want the lineup to get a second crack at.

Link to Bendix: Assessing the 2009 Tampa Bullpen