Rally-Killing & Dragon-Slaying, Quantified

1.) The Bridge – awarded anytime a reliever records three outs without allowing a run in a game separated by one run or less

Mike Adams (SDP) – 31/34 (91.2%)
Daniel Bard (BOS) – 28/32 (87.5%)
Luke Gregerson (SDP) – 26/33 (78.8%)
Rafael Betancourt (COL) – 23/28 (82.1%)
Joaquin Benoit (TBR) – 21/26 (80.8%)
Matt Thornton (CWS) – 19/21 (90.5%)
Ryan Madson (PHI) – 19/25 (76.0%)
Sean Marshall (CHC) – 18/22 (81.8%)
Sergio Romo (SFG) – 16/18 (88.9%)
Darren Oliver (TEX) – 15/15 (100%)
Joba Chamberlain (NYY) – 15/19 (78.9%)
Jonny Venters (ATL) – 13/21 (61.9%)

I would hazard that this is the situation we most associate with premium set-up men: lights-out reliever enters a very tight game in the seventh or eighth inning and gets his team back in the dugout with a chance to go for the jugular.  The rudimentary data suggests that success in this situation may be crucial to maintaining the manager’s trust.  Even guys like Betancourt and Chamberlain, who, as you may recall, were relatively inconsistent pitchers over the course of 2010, had high success rates in the “bridge” situation.

2.) The Dam – awarded anytime a reliever records four or more outs in a game separated by three runs or less

Darren Oliver (TEX) – 12/14 (85.7%)
Matt Thornton (CWS) – 11/12 (91.7%)
Daniel Bard (BOS) – 11/14 (78.6%)
Luke Gregerson (SDP) – 10/11 (90.9%)
Sean Marshall (CHC) – 9/11 (81.8%)
Jonny Venters (ATL) – 8/10 (80.0%)
Joba Chamberlain (NYY) – 7/9 (77.8%)
Sergio Romo (SFG) – 5/7 (71.4%)
Rafael Betancourt (COL) – 4/5 (80%)
Mike Adams (SDP) – 3/3 (100%)
Ryan Madson (PHI) – 3/4 (75.0%)
Joaquin Benoit (TBR) – 3/7 (42.9%)

While it seemed pretty standard for premium set-up men to convert Bridges at around a 80% clip and be hand the opportunity 20-30 times per season, their efficiency in Dam situations was all over the board.  In the case of the Padres duo, it was clear that manager, Bud Black, had utter confidence sending Gregerson back out for a second inning (or partial inning) of work, but no interest in doing the same with Adams, even though Black used them interchangeably much of the time.  Benoit, who excelled in nearly every other outing in 2010 (see 1.34 ERA), seemed to get creamed whenever he tried to get more than three outs.

3.) The Rallykiller – awarded anytime a reliever enters the game with runners in scoring position and succeeds in stranding them

Sean Marshall (CHC) – 12/20 (60.0%)
Daniel Bard (BOS) – 12/21 (57.1%)
Luke Gregerson (SDP) – 11/15 (73.3%)
Joba Chamberlain (NYY) – 10/15 (66.7%)
Darren Oliver (TEX) – 10/15 (66.7%)
Matt Thornton (CWS) – 9/12 (75.0%)
Joaquin Benoit (TBR) – 7/9 (77.8%)
Jonny Venters (ATL) – 7/11 (63.6%)
Rafael Betancourt (COL) – 6/14 (42.9%)
Sergio Romo (SFG) – 5/7 (71.4%)
Ryan Madson (PHI) – 2/4 (50.0%)
Mike Adams (SDP) – 0/3 (null)

To nobody’s great surprise, this is the hardest thing for a reliever to consistently accomplish.  I suspect anything better than a 50% success rate should be considered excellent.  Yankees fans are well aware of how good Joba has been in these scenarios over the past couple seasons and even he only managed to kill two out of every three rallies he encountered.  In these moments, as much as any other, he will be missed.  I found it very interesting that some guys who are generally considered top-flight set-up men, like Madson and Adams, are almost completely absolved of pitching with inherited runners.  This should probably be held against them somehow.

4.) The Dragonslayer – awarded anytime a reliever retires the opposing #3, #4, or #5 hitter at a time when a homer could result in a lead change (or tie)

Luke Gregerson (SDP) – 24/31 (77.4%)
Matt Thornton (CWS) – 42/56 (75.0%)
Sean Marshall (CHC) – 39/52 (75.0%)
Ryan Madson (PHI) – 27/36 (75.0%)
Darren Oliver (TEX) – 20/27 (74.1%)
Sergio Romo (SFG) – 19/26 (73.1%)
Mike Adams (SDP) – 35/48 (72.9%)
Joaquin Benoit (TBR) – 27/38 (71.1%)
Rafael Betancourt (COL) – 27/40 (67.5%)
Daniel Bard (BOS) – 43/66 (65.2%)
Jonny Venters (ATL) – 28/44 (63.6%)
Joba Chamberlain (NYY) – 22/37 (59.5%)

Clearly, opportunities for Dragonslaying are not evenly distributed.  In some cases, like Romo’s and Benoit’s, it seemed as though the manager was deliberately shielding his reliever from the meat of the order for substantial stretches of the season.  In others, like Oliver’s, it was difficult to see the rarity of chances as anything more than happenstance.  I do think it’s telling that the three youngest pitchers in this group – Bard, Chamberlain, and Venters – all struggled in these situations, although they were faced with them relatively often.

I’m not going to attempt any more rigorous analysis at this point, partially because the sample size remains small, but also because I hope the data itself will spur a little debate.

About Matt Seybold

Matt teaches at The University of Alabama. Roll Tide. He specializes in American Literature and Rhetorical Economics. Fate chose for him the peculiar perdition of rooting for the Chicago Cubs and the Los Angeles Clippers.

6 thoughts on “Rally-Killing & Dragon-Slaying, Quantified

  1. Interesting post. But I'd like to see more details about the Rally Killer stat. A reliever should get more credit for rally killing if he comes in with say a runner on 3rd and no outs as opposed to a runner on second and two outs.

  2. Thank you for the insight. I found your first article on the subject interesting and I appreciate the follow-up. I have a question regarding the conversion rates for #2 ( the Dam). How are the opportunities defined? If a reliever pitched an inning and then came back out and faced at least one hitter, does that qualify as an opportunity? I am assuming appearances of relievers who only pitch one inning and come out of the game do not count as an opportunity. Also, does the three runs or less separation rule apply throughout the four outs? For example, let's say a team is up three, a reliever gets one out to end the sixth inning, his team scores once in the top of the next frame, and then the same reliever gets all three outs in the seventh. Would this qualify as a Dam? Thanks for the clarification.

    • Thanks, Chris. It's a good question. As I proceeded, I certainly debated whether I would change what defined "opportunity" if I continued working on this. For the time being, however, I assumed to be on the hook for a "Dam" situation, the reliever had to face at least one batter after he had gotten three outs (thus giving him a shot at 4). So, you are correct, if the manager pulled his pitcher between innings, I wouldn't define that as an opportunity. What I was trying to see, after all, was which pitchers were trusted to go multiple frames. In terms of the score, if the pitcher entered a game that was separated by three runs or less, even if his team proceeded to extend the lead, I counted it. If the other team scored on him, of course, I also counted it. If he entered with a bigger differential and then the game was brought to within three runs, I did not count it. Such circumstances were exceedingly rare.

      In the case of the "Bridge," I also presumed that the pitcher had to be given the opportunity to get three outs. So, if his manager yanked a reliever after facing two hitters, I didn't count that as an opportunity. What this means is that there were several instances when a pitcher entered the game in a classic Bridge scenario, got torched, and didn't have it held against him. However, I figured if I did it in another way, I would end up with misleading "blown bridges," because pitchers were often pulled after successfully retiring one or two hitters, just because the manager wanted to exploit matchups by using a couple relievers in a single inning.

  3. I think some aspects of your first question are addressed in my response to Chris (written at the same time you were posting, I believe). I like your idea of a "pitches per out" stat for relievers. However, I think it's hard to judge how many outs a manager "would like" a pitcher to get. I'm guessing, in many cases, he doesn't know himself until he's seen that pitcher face a couple batters, which is part of why I have thusfar defined "chances" in this way.

    Fangraphs offers a version of what you're talking about, Shutdowns & Meltdowns, based on WPA, which I think is great. However, I wanted something that highlighted specific high-leverage situations we associate with quality relieving. My implicit thesis is that some relievers can excel in one type of scenario, but not in another, and it would be nice to know who is good at what. One of the things I noticed and which I might try to address at a later date, was that managers sometimes lost faith in their set-up men in one situation based on failures in another.

    • Thanks for responding…

      Thinking back, you are looking at least at "batters faced" which takes into account some of what I thought was missing. It is hard to numerically identify an opportunity; it may be the hardest part of the problem. You know there were times where they wanted Joba to go 5 outs, but after walking one and having a 12 pitch at bat he was unable to get more than the first 2. Even if he gave up no runs, it could be considered a bad outing. It would take a very complex set of rules to pick it out what an opportunity is.