The Closer and the Damage Doneby Steve Treder
August 17, 2004
Jerome Holtzman was a fine baseball writer for a very long time, working for several Chicago papers as well as for The Sporting News. His annual "Year in Review" pieces that appeared in the TSN Official Baseball Guide for many years were unfailingly thorough, generally well-balanced, and written in the straightforward, no-frills manner characteristic of a writer focused on the content of his work, and not much concerned with imprinting it with stylish flair or personality. Holtzman's body of work is impressive, and he well deserves the J.G. Taylor Spink Award granted him by the Baseball Hall of Fame.
But along the way Holtzman did one thing that had long-range consequences far beyond those of any other baseball writer of his time, perhaps of all time. These long-range consequences have been immense, and are still playing out, and are starkly visible on every MLB roster today, and in nearly every MLB game. In the opinion of this observer, the net impact of these consequences has not been positive.
In 1959, Holtzman was trying to come up with a means of measuring the impact and worth of the relief pitcher, a role that in the 1950s had come to be specialized and important as never before. Holtzman analyzed this phenomenon and invented a new statistic: the Save.
At first the Save stat was little more than an interesting toy, one of those fringe stats that folks have fun with all the time, like Game Winning RBI or Quality Starts. Broadcasters and fans may talk about them, but their fundamental flaws are pretty clearly recognized, and they aren't taken seriously by major league managers or GMs, and certainly don't have an influence on the way rosters are constructed or games are managed.
But the Save got increasing visibility as the 1960s unfolded, and the formula for determining Saves was tinkered with by Holtzman and others. By the end of the decade it had reached official status: the momentous original MacMillan Baseball Encyclopedia, published in 1969, included Saves, retroactively computed for all pitchers throughout history; and in 1970, TSN's Official Baseball Guide included the Save stat for the first time. The Save had arrived.
In 1977 24-year-old Bruce Sutter of the Cubs burst into prominence with one of the most stunningly brilliant seasons any relief pitcher has ever enjoyed. Into early August, Sutter had worked his drop-off-the-table splitter in 47 games and 86 innings, allowing just 49 hits and 15 walks, striking out 101, and posting a 5-1 won-lost record, 24 Saves, and a 1.06 ERA. Then he came down with back trouble, and sat out three weeks. He returned to pitch well in late August and September, but not quite as amazingly; his ERA over his last 22 innings was 2.42.
In 1978 Sutter came back strongly; through August 4th his ERA was 1.68 in 69 innings, with just 45 hits and 23 walks allowed, and 76 strikeouts. But then, apparently fatigued, he struggled mightily the rest of the way, posting a 6.44 ERA over his last 29 innings, surrendering 37 hits.
I distinctly recall reading the Sporting News interview with Cubs' manager Herman Franks in early 1979, reflecting on the late-season breakdowns of his young superstar fireman. Franks declared that from now on, in an effort to preserve Sutter's health, the split-fingered fastballer would rarely be used unless the Cubs were leading late in the game; he would predominantly be used, in essence, in Save situations. In 1977-78 Sutter had finished the game in 75% of his appearances, and recorded a Save 46% of the time. In 1979 he finished 56 of his 62 games (90%), and got the Save 60% of the time. A new method of deploying the ace reliever had arrived.
In 1985 Bill James published The Bill James Historical Baseball Abstract. One feature included in that book was "A History of Relief Pitching," (pp. 201-208), which presented by far the most thorough and perceptive overview yet seen of the role's development. James concluded his article presciently:
The question, for the future, is whether we have finally reached an equilibrium, or are we still in motion? I'm inclined to think we're still in motion. Save totals are now being pushed up year after year, and that means, to me, that the job is still subject to some redefinition. I think that if we look back at it again from 1995, we're going to be saying the same things then that we say now - that the role of the relief pitcher is different than it was a generation ago, that this is the first generation of "modern" relievers.
In 1987 Dennis Eckersley was a 32-year-old former ace starter whose career had seriously unraveled. Then he was converted to reliever by the Oakland A's, and his effectiveness blossomed anew. In 1988 manager Tony LaRussa, blessed with an extraordinarily deep and able 5-man bullpen, chose to deploy Eckersley primarily in one-inning Save situations (1.21 innings per appearance), and Eck responded with a terrific 45-Save performance. Never again in his career would Eckersley exceed 1.16 innings per appearance, and his great effectiveness for several seasons while being used in this extremely-specialized manner had tremendous influence: by the mid-1990s nearly every MLB team was deploying its ace reliever (whether a particularly effective pitcher or not) in a pattern very close to Eckersley's. The Closer had arrived.
In 2001 The New Bill James Historical Baseball Abstract came out, and James updated and expanded upon the continuing evolution he had foreseen. Three articles in particular address the issue:
- "Valuing Relievers" (pp. 232-239), in which James presents and assesses the five major models of relief ace usage since the 1930s (the Clint Brown pattern, the Elroy Face pattern, the Hoyt Wilhelm pattern, the Bruce Sutter pattern, and the Robb Nen pattern)
- "My Two Cents, 246 Words" (p. 295), which lays out five reasons why the modern use of left-handed short relief specialists (in the vernacular, "LOOGYs," for Left-handed One Out GuYs) isn't cost-beneficial
- A discussion of Eckersley's 1992 MVP selection (pp. 867-868), in which James equates awarding the MVP to a great short-relief specialist such as Eckersley to awarding an MVP to a great pinch-hitting specialist, such as Jerry Lynch in 1961.
The Robb Nen pattern that James discusses is Eckersley's. It's the modern Closer role with which we're all familiar: the relief ace deployed almost exclusively in Save situations, the vast majority of which are one-inning stints. This manner is different from the Bruce Sutter pattern primarily in its reliance on shorter stints: in 1979, for example, Sutter worked 1.63 innings per appearance, and modern aces in the Nen pattern are closely bunched around 1.00 to 1.15 innings per appearance.
As we saw last week and the week before, since the late 1980s, ace starting pitchers have been deployed in lighter seasonal workloads than ever before; top starting pitchers are generally throwing about 10% fewer pitches per season than they used to, meaning of course that other pitchers are by necessity throwing a higher proportion of each team's pitches than they used to.
A 10% reduction in annual pitch count workload is almost trivial in comparison to the degree in which the workloads of ace relievers have been reduced. Using Tangotiger's Pitch Count Estimator:
- The following is an incomplete list of relief aces who had at least one season in the 1960s, 70s, or 80s in which they threw an estimated 2,000 or more pitches: Ted Abernathy, Bill Campbell, Bill Caudill, Doug Corbett, Mark Eichhorn, Don Elston, Rollie Fingers, Terry Forster, Steve Foucault, Goose Gossage, Wayne Granger, Willie Hernandez, John Hiller, Tom Hume, Jim Kern, Bob Lee, Aurelio Lopez, Sparky Lyle, Mike Marshall, Lindy McDaniel, Greg Minton, Sid Monge, Ron Perranoski, Dick Radatz, Ken Sanders, Dan Spillner, Bob Stanley, Kent Tekulve, and John Wyatt.
- Knuckleballers Eddie Fisher, Charlie Hough, Hoyt Wilhelm, and Wilbur Wood all had seasons that far exceeded 2,000 estimated pitches.
- 1,500-to-1,800 estimated pitches per season was a typical workload of ace relievers, achieved by many pitchers annually, from the early 1950s until the late 1980s.
- The last full-time reliever to reach 2,000 estimated pitches in a season was Eichhorn in 1987.
- In his career as a Closer, Eckersley never threw as many as 1,200 estimated pitches in a season.
- Closers such as Eric Gagne, Trevor Hoffman, Mariano Rivera, and John Smoltz throw between 1,000 and 1,300 estimated pitches per season.
- The heaviest-worked reliever of the current era, Scott Sullivan from 1997-2001, was in the 1,500-1,800-estimated pitch window. The relief workloads of the other heaviest workers in recent years -- Octavio Dotel, Keith Foulke, Danny Graves, Shigetoshi Hasegawa, Byung-Hyun Kim, Billy Koch, and Derek Lowe -- have slightly exceeded 1,500 pitches.
We see that Closers in particular, and relievers in general, are handling workloads at least 25% or 30% lower than those commonly handled by relievers as recently as 15 or 20 years ago. Just as with starters, by necessity this means that lesser-accomplished, lower-quality relief pitchers are handling a significantly higher proportion of every team's workload than they were as recently as 15 or 20 years ago.
The deployment of the Robb Nen pattern/Closer model doesn't occur in a vacuum, of course; it's simply one key element in the systematic reconfiguration of pitching roles that's occurred over the past couple of decades. It's obviously interrelated with the simultaneous phenomenon of rapid near-extinction of the complete game (dropping from 13-15% in the mid-1980s to 4-5% today), and is only the most prominent of many examples of extreme specialization in relief roles, including Setup Man and LOOGY. Indeed to a certain extent the Closer can't practically exist alone; in order to deploy the relief ace as a Closer, it's pretty much a requirement for teams to specialize in their other bullpen roles, reserving all those ninth inning Save opportunities for the Closer.
This hyper-specialization in the bullpen has exacted large costs. In addition to the increasing proportion of relief innings handled by lesser pitchers, the size of bullpen rosters themselves has expanded in order to accommodate all the short-stint specialists. As James puts it in "My Two Cents, 246 Words":
... the manager is taking innings away from his sixth, seventh, and eighth best pitchers, and giving them to his tenth-, eleventh-, and twelfth-best pitchers. On almost all teams, this creates a demand for more pitchers than are available.
... to keep two extra pitchers, you have to cut a pinch hitter and a defensive player. Either player, if he is worth a crap, is worth more than five runs [the estimated benefit of gaining the platoon advantage 100 times a year by deploying LOOGYs]. If you pinch hit an outfielder for an infielder, you can get the platoon advantage, but you can also bring up a better hitter ...
It is inherently more difficult to control the platoon advantage on defense than it is on offense. Platooning at one position gives a team the platoon advantage an extra 200 times a season.
And what benefits have been gained by the modern expanded, specialized bullpen? Presumably, expectations were that two results would accrue: significantly reducing the workloads of star relievers would reduce their rates of injury (the motivation behind focusing Sutter's workload 25 years ago), and the more specialized leverage of reliever deployment, particularly ace reliever deployment, would yield more efficiency in the preserving of close leads, or converting Saves.
Have these benefits ensued? As in the case with starters that we reviewed last week, there is no evidence I'm aware of that reveals any meaningful reduction in injury occurrence rates among ace relievers. The following Closers are among those who have missed significant time due to injury within the past five years: Rod Beck, Joe Borowski, Tom Gordon, Eddie Guardado, Trevor Hoffman, Jason Isringhausen, Matt Mantei, Robb Nen, Troy Percival, Mariano Rivera, Kazuhiro Sasaki, Ugueth Urbina, Billy Wagner, Bob Wickman, and Jeff Zimmerman. Lots of ace relievers used in prior patterns also got hurt, of course, but the Robb Nen pattern doesn't appear to be proving to be particularly "safe" either.
Nor does particularly efficient production of Saves by modern specialized bullpens seem evident. Here are major league Saves per game, 1980-2003:
1980 21% 1981 22% 1982 22% 1983 23% 1984 24% 1985 23% 1986 24% 1987 23% 1988 25% 1989 25% 1990 26% 1991 27% 1992 26% 1993 26% 1994 24% 1995 25% 1996 25% 1997 25% 1998 26% 1999 25% 2000 24% 2001 25% 2002 25% 2003 25%The flatness of this line is especially worth noting when one factors in the great reduction in complete games that's occurred simultaneously; many victories completed by starters in the mid-to-late 1980s have since been handed to the bullpen to close out. Despite more opportunities than ever to convert Saves, they haven't materialized with meaningfully greater frequency. Strict specialization funnels nearly all of each team's total Saves into one Closer's bucket, but the total volume of Saves being created has remained about the same.
We have a bullpen model abiding today that costs teams at least one, often two or even three extra roster spots. It deploys the best relief pitchers far less liberally, and the worst relief pitchers far more liberally, than ever before. This bullpen model hasn't yielded any noticeable reduction in relief pitcher injury occurrence rates, nor has it delivered Saves with any noticeable improvement. What has caused its rapid and nearly universal adoption, and what causes it to persist?
I suspect there are two primary explanations behind the phenomenon: its benefits to managers, and its impact on contracts.
What's In It for Managers
The Closer model, with its highly specialized distinct bullpen roles, serves a purpose of greatly structuring and simplifying the in-game decision-making process for managers. Once the manager makes the determination of who his Closer is, who his primary Setup man is, who his LOOGYs are, and so on, then the decision of who to summon in various game situations becomes something close to following the recipe in a cookbook: when this happens, do that; once that's happened, do this. Pre-1979 Bruce Sutter might be brought in during a crucial spot in the 7th inning - a tricky tradeoff decision for his manager to weigh - but a manager following the Closer model faces no such challenge. No matter what, if it's the 7th inning, your Closer sits. One less thing to have to think about.
The tightly structured range of decisions also serves to shield the modern manager from a lot of second-guessing. The wide range of choices involved in deploying relievers outside of the Closer model exposes a manager to a wide range of choices that might work out badly, thus exposing him to pointed post-game questions from the press: why did you use this reliever then, or that way? But when each reliever is slotted into a tightly defined specialized role, and all the manager does is deploy each specialist within those bounds, then the manager rarely faces such tough questions. If you do everything "by the book," then the harsh light of criticism rarely finds you.
I'm fond of putting it this way: as a manager, if I use my relievers in a mode outside the current orthodoxy, and we lose, it's my fault, not theirs; but if I use my relievers the same way everyone else does, and we lose, it's their fault, not mine. Baseball managers are human beings who don't want to get fired, and as in all other walks of life, avoiding blame is a powerful motivator.
The Vicious Circle of Big Money
One of the most insidious impacts of the Save stat is the degree to which, by the 1990s, gaudy Save totals for ace relievers commanded huge bucks on the salary market. Joe Closer is able to take his 35 or 40 saves out on the free agent market and demand a king's ransom from teams eager to acquire a "Proven Closer." Once signed, Joe Closer's new team has little choice but to deploy him in the strict Robb Nen pattern - after all, that's what he's being paid all those big bucks to do. And if they don't, they'll be sure to hear about it from Joe and his agent, as the inflated market value of Saves creates a wicked jealousy between relievers as to who can win and keep the anointed Closer status - the sure ticket to riches. A team contemplating using its bullpen in a manner that won't offer a guaranteed 30+ Save total for the ace reliever knows it will encounter not only derision from the mainstream media, but also resistance from the relievers themselves, for very human reasons of economic self-interest.
Thus the Save stat created the Save Star, which created teams' insecure desire to have a Save Star, which created both teams' and players' interests in maintaining the existence of Save Stars. Once started, this engine has proven stubbornly unwilling to shut itself off.
Where Does This Leave Us?
So, is the Robb Nen pattern of bloated bullpens overstocked with mediocre specialists going to be with us forever?
Consider again what Bill James wrote in 1985:
The question, for the future, is whether we have finally reached an equilibrium, or are we still in motion? I'm inclined to think we're still in motion ... I think that if we look back at it again [ten years hence], we're going to be saying the same things then that we say now - that the role of the relief pitcher is different than it was a generation ago, that this is the first generation of "modern" relievers.
Those words proved to be true then. I think those words would have been true at virtually any moment in the 20th century. I think those words are still true today.
Every previous model of pitching staff/bullpen deployment offered strengths, but weaknesses as well. Sooner or later, the weaknesses came to be seen as intolerable by one or more teams. Someone eventually had success with a revised model, competitive dynamics stimulated others to follow, and soon a new paradigm had taken hold. Then, sooner or later, the weaknesses of the new paradigm would become apparent.
I see no reason to expect anything different regarding the current situation. Already, within the past few years, we've seen a few GMs openly questioning the wisdom of the Robb Nen model and initiating (with limited conviction and success, to be sure) some deviation from it. I think it's just a matter of time before someone stumbles upon a significantly new arrangement (possibly out of desperation), and it will do well. A few others will feel the need to copy it in some fashion, and soon the tipping point toward the new paradigm will be at hand.
Think of it this way: throughout history, every time anyone has said, "We've reached the endpoint of progress. The way things are is the way they will remain" - subsequent events have proven them wrong. I can't say for certain when and how the Robb Nen pattern of bullpen hyper-specialization will go the way of all previous patterns. But I have great confidence in saying it will.
References and Resources
The in-season stats for Bruce Sutter were found on that marvelous resource, Retrosheet.
Steve Treder can often be found spending way too much time talking baseball at Baseball Primer. He welcomes your questions and comments via e-mail.