:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/26527757/20130512_jrc_bh1_011.0.jpg)
The strikeout is the tonic that relieves all ills faced by a pitcher, regardless of situation. There are situations when a groundball can yield a better outcome, however (a double play), raising several questions about the true value of relief pitchers who can more reliably achieve that outcome.
To help arrive at some answers to those questions, I pulled the top ten seasons of relief pitchers in the last two years, according to ground ball rate (min 50 IP, with ranks among 134 RPs in 2012 and among 125 RPs in 2013):
| Name (Year) | IP (G) | GB% (Rank) | FIP | ERA | FIP-ERA | fWAR |
|---|---|---|---|---|---|---|
| Brad Ziegler (2012) | 68.2 (77) | 75.5% (1st) | 3.21 | 2.49 | 0.72 | 0.7 |
| Brad Ziegler (2013) | 73 (78) | 70.4% (1st) | 3.40 | 2.22 | 1.18 | 0.6 |
| Seth Maness (2013) | 62 (66) | 68.4% (2nd) | 3.43 | 2.32 | 1.11 | 0.1 |
| Jamey Wright (2012) | 67.2 (66) | 67.3% (2nd) | 3.39 | 3.72 | -0.33 | 0.3 |
| Eric O'Flaherty (2012) | 57.1 (64) | 66.0% (3rd) | 3.27 | 1.73 | 1.54 | 0.6 |
| Ronald Belisario (2012) | 71 (68) | 64.5% (4th) | 3.09 | 2.54 | 0.53 | 0.7 |
| Pedro Strop (2012) | 66.1 (70) | 64.3% (5th) | 3.59 | 2.44 | 1.15 | 0.7 |
| Matt Albers (2013) | 63 (56) | 63.8% (3rd) | 3.49 | 3.14 | 0.35 | 0.3 |
| Chad Qualls (2013) | 62 (66) | 63.3% (4th) | 3.32 | 2.61 | 0.71 | 0.5 |
| Jonny Venters (2012) | 58.2 (66) | 62.8% (6th) | 3.76 | 3.22 | 0.54 | 0.1 |
A first thing to note about these ten seasons is the large discrepancy between ERA and FIP (average of 0.75), which is consistently larger than the discrepancy for all relievers in 2013 (FIP – ERA = 0.11) or 2012 (FIP – ERA = 0.12). There are at least three likely reasons for this.
First, if these relievers tended to be brought in mid-inning more than the average reliever, they would reap that benefit disproportionately; that’s one reason to consider shifting the innings credit for outs toward the first out of the inning. Second, to the extent that these relievers tended to start outings with runners on base (and in 2013, 43.40% of all base-out states included a runner, and 14.05% included at least 2), it was easier for them to record more overall outs. Runners mean more potential ways to record outs, and the potential for multiple outs. Third, it’s possible that these relievers -- as opposed to relievers overall -- were particularly adept at recording double plays when they inherited a situation in which that was possible.
If elite GB% relievers are undervalued by the market, it could be due, in part, to reliability. It may be that for these pitchers, their success as measured by FIP and its components are more stable. Restricting the pitching seasons in the Steve Staude correlation tool to seasons of 50-80 IP (n=539), GB% shows a stronger correlation from season to next season (0.779) than K/9 (0.702). That’s a good thing, as GB% understandably has some bearing on same-season HR/TBF (-0.370, n=1111) and next-season HR/TBF (-0.324).
It could also be due, however, to being particularly well suited to a situation that comes up with some frequency (the third of the possible low-ERA explanations above), similar to how some relievers enjoy large boosts to their effectiveness through consistent use in arm-side matchups. I'm referring to double plays, although there is some overlap between platoon splits and DP conversion rate (there’s more than one RP in our group who throws with a low arm angle). Some of what we know about arm-side specialists can be applied here, but note that there are significant differences between taking advantage of arm-side matchups and taking advantage of relievers adept at handling certain base-out states. For one, platoon (and other batter-hitter) matchups are about the player at the plate. Largely as a result, a glance at the statistics of a LOOGY are unlikely to understate the effectiveness he may have if the batters he faced followed average distributions.
Rate stats like FIP can help us judge whether our elite GB% relievers are good pitchers (although ERA could clearly be misleading). But FIP and other DIPS-based formulae will not help us determine whether those pitchers are particularly adept at recording double plays in double play opportunities.
WPA/LI and RE24 can help, particularly the latter, as we work to determine how much of our elite GB% relievers’ success was due to the situations they were put in, and how much was due to being particularly good at handling those situations.
| Name (Year) | IP (G) | GB% (Rank) | WPA/LI (Rank) | RE24 (Rank) | FIP (Rank) |
|---|---|---|---|---|---|
| Brad Ziegler (2012) | 68.2 (77) | 75.5% (1st) | 1.34 (13th) | 20.24 (4th) | 3.21 (47th) |
| Brad Ziegler (2013) | 73 (78) | 70.4% (1st) | 1.64 (7th) | 14.22 (17th) | 3.40 (T-64th) |
| Seth Maness (2013) | 62 (66) | 68.4% (2nd) | 0.44 (T-71st) | 14.76 (16th) | 3.43 (69th) |
| Jamey Wright (2012) | 67.2 (66) | 67.3% (2nd) | -0.15 (T-108th) | -6.40 (123rd) | 3.39 (T-63rd) |
| Eric O'Flaherty (2012) | 57.1 (64) | 66.0% (3rd) | 1.08 (T-28th) | 16.03 (10th) | 3.27 (53rd) |
| Ronald Belisario (2012) | 71 (68) | 64.5% (4th) | 1.44 (10th) | 7.84 (57th) | 3.09 (T-37th) |
| Pedro Strop (2012) | 66.1 (70) | 64.3% (5th) | 0.35 (T-80th) | 12.10 (29th) | 3.59 (T-77th) |
| Matt Albers (2013) | 63 (56) | 63.8% (3rd) | 0.14 (T-92nd) | 0.13 (102nd) | 3.49 (74th) |
| Chad Qualls (2013) | 62 (66) | 63.3% (4th) | 0.39 (75th) | 7.65 (52nd) | 3.32 (T-61st) |
| Jonny Venters (2012) | 58.2 (66) | 62.8% (6th) | -0.33 (T-115th) | 1.37 (100th) | 3.76 (91st) |
Generally, our elite GB% relievers do better on these metrics than FIP might have led us to predict. On average, the ten players jumped almost thirteen places in rank from FIP to RE24, suggesting that there's something good about elite GB% that is not captured by FIP.
I have my doubts about whether further study through WPA/LI or RE24 could be helpful, in part because I’m not sure if the phenomenon I’m trying to look at is restricted to just the very best GB% RPs, in part because both WPA/LI and RE24 still depend so much on how RPs were deployed, and in part because DPs are not the only thing that might cause higher-than-expected RE24 results. Because a strikeout is still clearly superior to a ground ball or even a double play in some circumstances, for example (a study could be done on strikeout percentages and the scoring rates of inherited runners, perhaps), high-K% RPs could consistently outperform an expected RE24 if they are frequently deployed in situations in which a strikeout was particularly helpful.
Must Reads
In terms of DPs, however, we can do better – by actually looking at DPs. RE24 has some answers for us, but the table above says nothing definitive about whether the ten RPs were actually used in DP situations at all. In double play opportunities (first, first and second, first and third, loaded; zero outs or one out), double plays were recorded just less than 13% of the time in the relevant seasons (12.95% in 2012, 12.90% in 2013). As we might expect, most of our elite GB% relievers outperformed league average in recording double plays on a rate basis:
| Name (Year) | IP (G) | GB% (Rank) | DP | DP Opp | DP/DP Opp |
|---|---|---|---|---|---|
| Brad Ziegler (2012) | 68.2 (77) | 75.5% (1st) | 21 | 70 | 29.58% |
| Brad Ziegler (2013) | 73 (78) | 70.4% (1st) | 8 | 58 | 13.79% |
| Seth Maness (2013) | 62 (66) | 68.4% (2nd) | 16 | 54 | 29.63% |
| Jamey Wright (2012) | 67.2 (66) | 67.3% (2nd) | 5 | 58 | 8.62% |
| Eric O'Flaherty (2012) | 57.1 (64) | 66.0% (3rd) | 11 | 51 | 21.57% |
| Ronald Belisario (2012) | 71 (68) | 64.5% (4th) | 7 | 49 | 14.29% |
| Pedro Strop (2012) | 66.1 (70) | 64.3% (5th) | 12 | 63 | 19.05% |
| Matt Albers (2013) | 63 (56) | 63.8% (3rd) | 10 | 57 | 17.54% |
| Chad Qualls (2013) | 62 (66) | 63.3% (4th) | 11 | 47 | 23.40% |
| Jonny Venters (2012) | 58.2 (66) | 62.8% (6th) | 10 | 51 | 19.61% |
All but one of the ten in our group outperformed league average, and as a group, they converted DPs at a rate of 19.89%, or almost exactly 7% above league average.
The average GB% for relievers in 2013 was 45.3%, and there were three RPs in 2013 with more than 50 IP to record a GB% of 45.3%. Of these three, each recorded a DP/DP Opp rate of around the leave average: John Axford (10/100, 10%), Troy Patton (8/51, 15.7%), and Alfredo Simon (5/65, 7.7%). Widening the field to two tenths off of the mean, we add Anthony Swarzak (45.2%, 11/69, 15.9%), Edward Mujica (45.2%, 6/36, 16.7%), Robbie Ross (45.4%, 7/71, 9.9%), Luke Gregerson (45.5%, 1/39, 2.6%), and Logan Ondrusek (45.5%, 3/43, 7.0%). Together, those eight pitchers got 51 DPs in 474 opportunities (10.8%), which is a lot closer to the league average conversion rate than our elite GB% guys.
So on to the big question: what kind of "extra" value do elite GB% RPs have in converting DPs more frequently than other pitchers?
As a first step in answering that question, we can help to translate frequency to occurrence by figuring out how often DP situations actually arise. The eight base-out states that we've defined as DP opportunities occurred 33,778 times in 2013. The average team, therefore, had about 1126 DP opportunities. Since it's very unlikely that elite GB% RPs will get deployed in early innings or in ninth innings, I think we could divide by three to narrow down to DP opportunities in innings six through eight (this is not meant to be exact, anyway but it bears noting that DP opportunities might be slightly more prevalent in those innings and in first innings, based on scoring). That leaves about 375 DP opportunities -- but since managers are probably loathe to routinely take out an RP that just entered the game to bring in a specialist for a DP opportunity, we should reduce that number somewhat. In 2013, the average relief appearance was almost exactly one inning, and with four-to-five base-out states in most innings, I think we can cut out about half of our number, taking 375 to about 188.
I may have cut that down a bit aggressively, but in terms of how often an elite GB% RP might be called on to pitch, we can cut it down further -- there's little reason to deploy him in a blowout. The nature of that last adjustment is a matter of taste, but it's not nothing. I'm thinking close to half -- taking my 188 number to 100 or so.
Those 100 DP opportunities could happen multiple times in a game, or four days in a row, etc. It's unrealistic to think that a single elite GB% RP could cover all of those -- and so it makes sense that the DP opportunities in the table above top out in the 60 range.
As an aside, this innings issue is part of what makes this so interesting to me. With the possibility that single elite GB% RPs could end up handling a strong majority of the highest-leverage DP opportunities, it would make little sense for a team to hoard more than one of these elite GB% RPs.
If elite GB% RPs have a chance of converting a DP opportunity that's about 7% higher than league average, then over 60 opportunities, we'd expect to tally about 4 additional DPs. Our DPs have not been GB-specific (lineouts do happen), but since we're talking about GB relievers, I'd be most comfortable assigning a -0.85 run value to those 4 DPs. That's 3.4 runs -- good for about 0.3 wins. If a freak like Brad Ziegler or Seth Maness could demonstrate a real ability to convert DPs at a level above the 20% mark -- let's say, 25% -- that could make a real difference, but only up to about 0.5 wins.
Not all outs are made equal, and the Leverage Index for our eight base-out states in innings six through eight do make wonder if the "extra" runs saved have a significantly disproportionate impact on actually winning individual games. The WPA/LI for our 10 RPs are so varied, however, that I conclude that the usage of each of our 10 RPs was also very varied. So varied, in fact, that gaining some clarity on whether our DP wins affect games disproportionately would require a completely different (prospective) exercise.
So back to the original question: are elite GB% relievers undervalued? In terms of FIP, and by extension WAR, I think the answer is yes. But given that REW seems to more or less match the sum of WAR and the DP-dependent value that's missing from WAR, I think it's likely that RE24 values elite GB% relievers properly.
In terms of dollars, I'm not so sure -- there are forces pulling in both directions. FIP may not properly reward elite GB% relievers for the extra value they can generate through double plays -- in some respects, FIP may be "artificially low" for that select handful of relief pitchers. But it's not like teams don't understand that GB% RPs can add value by getting the team more DPs -- that's precisely why they get deployed in DP opportunities in the first place. And to bring this article full circle, ERA appears to be a determinant of arbitration salaries, and ERA is definitely artificially low for elite GB% types who tend to get brought in with men on base.
. . .
Statistics courtesy of FanGraphs. Many thanks to John Choiniere (@JohnChoiniere) for his invaluable research assistance.
Ryan P. Morrison is a contributor to Beyond the Box Score, and co-author of Inside the 'Zona, a blog on the Arizona Diamondbacks. You can follow him on Twitter: @InsidetheZona.
