Earlier this week, in a story that was appropriately reported without much fanfare,Eric Gagne retired. Between 2002-2004, Gagne had been one of the best relievers in the game saving 152 games and posting a 1.79 ERA over that three year period, while appearing in three All Star Games and finishing 4th, 1st and 7th in Cy Young balloting. For those three years, Gagne was the best closer in baseball, but he never recovered from an injury in 2005 and bounced around between the Dodgers, Rangers, Red Sox and Brewers between 2005 and 2008. He spent last year trying to get one more chance to play, but was not successful.
Gagne’s story is an unfortunate one, but it is interesting because career paths like his, a few great years, or even one great year before succumbing to injury, obscurity or bad luck are far more common among closers than among either starting pitchers or hitters. In general historically great starting pitching and hitting seasons are produced by players who also produce great careers, but this is not true for closers. Gagne is one of several closers who have contributed one of the best years ever by a closer but will not get anywhere near, for example, ever being elected to the Hall of Fame.
One way to see this is to identify the best seasons ever by relievers and compare them to the best seasons by hitters and starters. For relievers, I defined best seasons as ones where a pitcher saved 40 or more games while allowing fewer than one hit plus walk per inning while posting an ERA+ of 250 or better. For starting pitchers the criteria were 15 wins, one or fewer hits plus walks per inning and an ERA+ of 150 or better. For batters, I simply took the top 25 seasons according to OPS+. However, because all of the relief seasons that met this criteria occurred after 1990, I only looked at seasons since 1990 for the other players as well. Had I not been using the 1990 cutoff, I would have raised the win criteria to 20 rather than 15.
Of the sixteen seasons who met this criteria, all but four, one each by Kevin Brown, Jason Schmidt, Kevin Millwood and Derek Lowe were produced by likely Hall of Famers such as Greg Maddux or players who are still active and will be strong candidates such as Johan Santana.
The story is somewhat different for the fifteen relievers in this group which include Hall of Famers such as Dennis Eckersley, active pitchers such as Mariano Rivera and Trevor Hoffman who are strong candidates for the Hall of Fame, but also several pitchers such as Eric Gagne, Armando Benitez, JJ Putz, Michael Jackson, Rob Nenn and Bryan Harvey who combine for seven of the fifteen seasons on the list, who will almost certainly not be remembered as all time greats.
Among hitters, the top 20 OPS+ seasons since 1990 were achieved by players who are either very strong Hall of Fame candidates such as Albert Pujols and Frank Thomas or players whose candidacies are tarnished by steroid issues but who otherwise would be easy Hall of Fame selections such as Barry Bonds. Only one player on the list, Jason Giambi failed to sustain that greatness over an entire career. In general, the data suggests that great seasons by less than great players are more common for pitchers than for hitters, and more common for starters than relievers.
There are several explanations for this. First, there is a strong subjective element to my approach. The criteria were chosen more or less off the top of my head as were notions that players like Jonathan Papelbon or Johan Santana might become historically great pitchers, while JJ Putz, for example, will not. Nonetheless, this data provides mild quantitative support for a notion that seems anecdotally to be true. The basic findings do not change when the criteria are changed slightly or when a greater time frame is used. In some cases, such as when ERA+ is the only measure, the findings are more stark.
Every year, it seems like a few top closers emerge while every year a few fade out. In the last quarter century or so, several starting pitchers such as Maddux, Randy Johnson, Roger Clemens and Pedro Martinez have dominated the game, while only two relievers, Rivera and Hoffman, have approached this level of consistency. Even standout seasons by less than historically great starting pitchers can be explained away by injury (Dwigh Gooden) or the player ended up producing a very good if not great career (Ron Guidry, Vida Blue). While Gagne’s decline can be explained by injury, and Nen or Benitez might be comparable to Guidry, Putz, Harvey and others, including seasons like Jim Kern’s 1979 season, are just cases of relievers having short and unpredictable peaks.
Another explanation is that we lack good and broadly acknowledged measurements for closers. ERA and related statistics were developed for starters. Saves, like its cousin wins, is very dependent on how the team plays and how the pitcher is used. Additionally, because modern closers pitch so few innings a year, a few bad outings can skew a pitcher’s numbers for a whole season.
These explain some but not all of what may cause this phenomenon. Other possible explanations include that the nature of being a closer may be more susceptible to pressure or may allow pitchers with one great pitch to excel for brief periods of time-although Mariano Rivera has done this over a lengthy career. Regardless of possible explanations, there are some clear lessons from this data, all of which reinforce the absurdity of the notion of the “established closer”. First, signing closers to long term contracts is generally an even bigger risk than signing starting pitchers to a long term contract. Second, good closers can be found by looking in unexpected places. Third, once a closer has a bad year, it is probably time to let him go. These are just preliminary findings which will undoubtedly need to be modified once we develop better measures specifically for closers, but they are still a useful beginning.