Skip to main content
Original Issue

Undiscovered MODEL


IN 2005, four years removed from his major league catching career, Scott Servais took a job as a Rockies scout. His boss and former teammate, Jerry Dipoto, Colorado's director of player development, was a former reliever with a reputation as something of a wonk—at the start of games Dipoto would pose a trivia question to his teammates in the bullpen—and as a young executive he had taken to statistical analysis as the Moneyball revolution was taking off.

Dipoto, who was a scout on the 2004 Red Sox championship team, had a mantra inspired by Bill James: Do the work yourself so you can be truly sure. Servais found out just how seriously Dipoto took this maxim as they ran studies on everything from the career arcs of Latin American players ("They sign earlier but don't get to the big leagues earlier," says Servais) to the similarities between winning major league franchises. ("The secret is clearly in player development," says Dipoto.) For one project that delved into the correlation between age and prospect projection—they found that for nine out of 10 prospects, you can gauge potential outcomes simply by focusing on their age relative to their level—Servais combed through every page of the Baseball Encyclopedia to scrape data on players with 2,500 at bats and starters who threw 500 innings. The study took 2½ years.

One baseball mystery, though, remained elusive. The year-to-year volatility in the performance of relief pitchers has long confounded teams, and after countless bullpen studies, Dipoto found himself coming back to one conclusion: "The idea that you're ever going to have a surefire way to create a bullpen is simply preposterous."

A decade later, Dipoto (who became general manager of the Mariners last September) and Servais (whom Dipoto picked as Seattle's new manager a month later) have reunited and are trying to turn around a franchise with the longest active playoff drought in the game. Dipoto overhauled the team—between Oct. 19 and Opening Day, he moved 58 players on or off the 40-man roster—and aggressively addressed the team's biggest problem: the bullpen, which had collapsed to the sixth-worst in baseball in 2015 after leading the majors in ERA the year before.

Dipoto did something bold for an era in which relievers have become more important and costlier than ever: He constructed an almost entirely new bullpen on the cheap, with a group of mostly unheralded, buy-low relievers. Among the Mariners' new relievers are a whippy sidearmer (closer Steve Cishek), a 38-year-old with a traditional overhand delivery (Joaquin Benoit), a big power arm with a Wiffle-ball slider (Tony Zych), a cutter-centric righthander (Nick Vincent) and a pair of righthanders whose fastballs top out in the high 80s but have still racked up swings and misses (Joel Peralta and Steve Johnson).

Nearly a third of the way into the season, the Mariners, who were in first place in the AL West through Sunday, have had one of the best bullpens in all of baseball (with a 2.63 ERA, ranked fifth in the majors). Meanwhile, teams that invested unprecedented amounts in relief pitching over the winter were stumbling. Servais's anonymous, soft-tossing relief staff had the look and feel of an overachieving unit enjoying a fluky first two months. Or could it be a new model of how to build an economical bullpen in 2016? Dipoto and Servais are tackling one of the most confounding enigmas in baseball.

LIKE ALL sports, baseball is a game of copycats: One team figures out a better way of doing something, and others follow. The success of Tony Gonzalez and Antonio Gates started a trend of NFL teams drafting basketball-playing tight ends. The Spurs won with radical resting strategies, and other NBA teams tried to do the same. After the Royals won back-to-back American League titles and a World Series last fall on the back of a historically good bullpen, it was inevitable that other teams would imitate their championship blueprint and try to overwhelm opposing teams with dominant relievers.

"The way the Royals did it—you had to take note," says Dipoto, who was GM of the 2014 Angels team which won an MLB-best 98 games but was swept by the Royals in the Division Series. "When you got into that bullpen, you knew that you were done because they're going to run out nine straight outs and you'll have a difficult time just fouling it off. That's a dejecting feeling."

Says Rangers assistant GM Thad Levine, "The perception that you could have a good but not great starting rotation and invest, relatively speaking, significantly less money on exceptional relievers in relation to good starters, and win that way: That led teams on a path they'd not gone down in recent years. In the past, teams would start conversations asking what starters would be available, and now the first thing they would ask you about was your setup man."

This past off-season saw a flurry of trades and signings confirming that shutdown relievers are more coveted than ever. Teams retained their 2015 closers yet still paid a premium for an additional elite arm: The Yankees (Aroldis Chapman), Red Sox (Craig Kimbrel), Astros (Ken Giles), Blue Jays (Drew Storen), Rangers (Tom Wilhelmsen) and Royals (Joakim Soria) all had closers in place but still added a pitcher who either finished last season as his team's closer or led his team in saves.

With some of the most data-driven organizations taking this approach, the old sabermetric tenet about bullpens—Don't pay big for relievers because they're unreliable and effectively interchangeable—was lit on fire. The Astros dealt away five players (including 23-year-old starter Vincent Velasquez and the No. 1 overall pick from 2013, starter Mark Appel) to get Giles, the Phillies' closer last year. The prices teams paid for setup men was nearly as jaw-dropping: The Cubs signed Trevor Cahill, who had been released in June, for $4.25 million; the Orioles re-signed Darren O'Day for $31 million over four years; the small-market A's shelled out for John Axford ($10 million over two years) as well as Ryan Madson (a three-year, $22 million deal).

"Billy is as interested in buying movable assets as anything—to take a high-demand player who he'll flip to get high-value, long-term assets," says a National League GM of Oakland executive VP Billy Beane. "So [the Madson and Axford deals] were a validation of this trend: They saw the market and acted accordingly."

This was not merely the Royals effect: Declining offense has put a premium on run prevention, and teams are adapting to the high cost of starting pitching. Many managers, swayed by statistical evidence that starters facing a lineup for the third time in a game perform dramatically worse, are also relying on their relievers more than ever: Starting pitchers logged more starts of less than six innings last season than in any other.

The problem with sports trends: Teams will overreact to what worked and what seemed innovative, and go so far in the opposite direction that the strategy becomes unprofitable. For now, at least, it looks like the Astros bought high on Giles, who in '14 and '15 was one of the game's most dominant relievers, with a combined 1.56 ERA. Giles had a 6.23 ERA through Sunday and allowed more home runs (four) over the season's first seven weeks than he did over the previous two seasons combined (three). Meanwhile, Velasquez has been one of the season's breakout stars or Philadelphia.

"There's a greater risk associated with relievers—the predictive measures are less reliable," says Vince Gennaro, president of the Society for American Baseball Research and an analytics consultant to MLB teams. "Conceptually, it should be getting easier now that we have new tools: spin rate and pitch movement, which may be able to correlate with success. But relievers will always be far less predictable than starting pitchers, simply because of the amount of work they get. An average reliever is going to pitch roughly a third as much as a starting pitcher. Evaluating a reliever on a season's results is the equivalent of evaluating a [starter] on two months' work. There's a case to be made that if a reliever has a high plateau and has come off that, it could be that something is going to regress back from that high level."

The AL West is Exhibit A in the volatility of bullpens, and not just because of Seattle's dramatic improvement. In the second half of last year, the Rangers' bullpen helped carry Texas to a surprising division crown. Texas went into the season with virtually that same unit in place—and results have been markedly different: Through Sunday the team had lost a major-league-high six times in walk-off fashion, and ranked last in the AL in bullpen ERA: 5.43. Last week Shawn Tolleson, who saved 35 games last year, lost the closer job after blowing his fourth save. "This off-season we actually felt like we had a good bullpen," says Levine. "And here we are, thinking we may need to address it again. But we can't lose sight of our strategic perspective: Outside of the Joe Nathan commitment we made [a $14.75 million two-year deal in 2012], we haven't allocated a major multiyear deal to a relief pitcher. By and large we've been more on the side of selling high on relievers, rather than investing at the height, just because of the volatility."

Some in the industry believe that relievers are actually more volatile now for the same reasons they are more dominant than ever. "There are so many more who are wired for the short-burst, high velocity," says Dipoto. "We're really exploring new areas with bullpens that have never been explored before. What I might have learned 15 years ago in researching bullpens probably isn't applicable today because of how the game has changed, because of how the back-end reliever has evolved. There's so much velocity, so many specialty pitchers, so much nuance that we have to adapt to."

SEATTLE'S BULLPEN is certainly not the game's most intimidating, but it may be the most interesting. Through the early part of the season it was one of the AL's softest-tossing units, yet it was still near the top in strikeouts. In his three years with the Angels, Dipoto earned a reputation for building strong bullpens on the fly, and he has taken his philosophy to Seattle: "It's trying to create the deepest group you can create. You want different options, you want different looks. [This winter] while everyone else was focused on guys throwing 100, we went to the point in the market where we found value in the guys who threw 90 instead of 100 and did it in a unique way."

New tracking technology has allowed analytics-driven organizations like the Mariners to be more creative in their player evaluations. For instance, in the cases of Peralta and Johnson, the Mariners looked past velocity and saw the exceptional spin rates on their fastballs, using the TrackMan video system that has been installed at some ballparks over the last few years. Spin rate—the number of times the baseball rotates on the way to the plate—has become valuable in player evaluation: Research shows that average fastball spin has a higher correlation to swinging strike rate than average fastball velocity. "Johnson's fastball is 88, and somehow they miss it," says Servais, who receives a report on spin rates every week.

The unpredictability of a player like Johnson—a lightly regarded reliever who was designated for assignment by the Orioles over the off-season and released during spring training by the Rangers before the Mariners signed him in March—is also the reason Texas hopes that things will soon turn around for Tolleson and the rest of their bullpen. And it's why Dipoto and Servais know that as good as their relievers have been so far, they are in for their own struggles. The pair may look like they have the answers one day, but the next, everything can change—that is the mysterious nature of bullpens.

For now, Dipoto is certain only of this: "If you think you've got it figured out, you probably don't."

Bullpens are volatile—just ask the teams who've seen the biggest swings in reliever ERA from last year to this year.


2015 ERA


2016 ERA



2015 ERA


2016 ERA



2015 ERA


2016 ERA



2015 ERA


2016 ERA



Photograph by Simon Bruty

NO BULL With high-velocity relievers more prized than ever, the Mariners went another way, signing a varied group of hurlers like sidearming closer Cishek.



TEXAS STRANGERS While Dipoto, a former reliever himself (top left), overhauled Seattle's pen, the Rangers kept a successful group intact—and have seen drop-offs from, among others, now-demoted closer Tolleson (above).



[See caption above]



[See caption above]



[See caption above]