Fifteen years ago, there were still MLB teams using 10-man pitching staffs for significant stretches of the season. A decade ago, just about everyone was using 11. Five years ago, 12 became the norm, and in 2013, for the first time, as many teams consistently carried 13 pitchers on their 25-man roster as carried any fewer.

There’s more than one reason for this. Starting pitchers pitch less deep into games than they once did, and teams are beginning to concede that fact and work around it, rather than push hurlers past their limits. Relievers are also being used differently. Most average three or fewer outs per appearance, and are asked to rear back with 15 of the nastiest pitches they can muster, rather than to work the team out of jams and act as bridges when starters have rough nights.

All over the league, the per-game workload of relief pitchers is dwindling, almost to nothing, almost to irrelevance. A 25-percent strikeout rate and 3:1 strikeout-to-walk ratio are the replacement level for these one-inning (sometimes one-out) short-shift workers. Offense is dying because these relievers are taking over, but not because of their dominance. Rather, it’s because teams have forced themselves to expend so many resources on accumulating tons of these arms, and on carrying so many of them at all times, and on compensating for the inevitable mistakes they make in the course of building that staff.

A small group of relievers, though, is trying to make leveraged long relief sexy again, and if not sexy, then at least viable. Craig Stammen of the Washington Nationals is one such guy. The Atlanta Braves, to their credit, have found success with a few guys in that role. Of the long relievers left in MLB, though, there can be no challenge to Josh Collmenter’s throne. The Arizona Diamondbacks signed Collmenter to a two-year, $2.425-million deal last week, marking the next chapter in the most interesting pitching story in baseball today.

Collmenter was a non-prospect at the beginning of 2011, short on stuff, long on deception, but seeing his numbers slide backward with each forward step he took toward the big leagues. His entry in Baseball Prospectus 2011 read, in part, “He might project as a fifth starter in the big leagues… But probably not.”

Yet, he spent nearly that entire 2011 season as a fifth starter for Arizona, and that team won the NL West, in part thanks to Collmenter’s 3.38 ERA in over 150 innings. It was a huge developmental win for the organization, turning a pitcher who only occasionally touched 90 miles per hour into a solid role player on a playoff team. If Collmenter had never thrown another pitch in the big leagues, 2011 would have easily justified the 15th-round pick the Diamondbacks had spent on him, and then some.

Before we go on, I want to stop and explain about Collmenter. I’ve alluded, already, to the deception he offers batters, and to the lack of great stuff or fastball velocity that made scouts so dubious of him. The thing I haven’t really mentioned, though, is the way Collmenter throws. His bizarre delivery is as much a part of his story as his arsenal, or his role, or his journey. See, Collmenter throws over the top. I don’t mean that he has a high arm slot, the way many pitchers used to, or that he kicks especially high and tries to really rocket the ball downward, toward the batter. I mean that Collmenter’s pitches seem to come out of the top of his head. His arm slot is so radical that, despite tepid stuff, he keeps hitters very uncertain, and makes it exceptionally difficult to pick up the ball. In pitching-mechanics parlance, arm angles are often discussed as numbers on the clock. A true side-armer throws from nine o’clock. Most guys are between 10 and 11 o’clock. Collmenter might throw from one o’clock; his delivery is that peculiar.

As you might imagine, big-league hitters, being very, very good, don’t just throw up their hands forever when a pitcher comes at them from a strange angle. Collmenter came back in 2012 as a starter, but after four very rough starts (he had a 9.82 ERA), the Diamondbacks sent him to the bullpen, instead.

That might have been the end of the story, or at least the interesting part, for a lot of guys. Relievers are fungible. They pitch so little, and in such selective situations, that they are neither reliable nor irreplaceable. Fundamentally, the one-inning reliever is a non-asset, unless he posts goofy strikeout numbers for at least two consecutive seasons.

Collmenter, though, didn’t become a one-inning reliever. He became something more. The Diamondbacks found that, although batters figured out Collmenter’s delivery eventually and were more than able to handle his raw stuff once they did, the first time through the order or so remained a breeze. He’d be pressed back into starting work briefly, later in the year, but Collmenter found his niche the day he became a long man. He threw 31 innings in 17 relief appearances in 2012, and ended up making 11 starts. His final ERA was 3.69, an impressive turnaround after his disastrous April, and he backed that up with a 4:1 ratio of strikeouts to unintentional walks.

Then, he topped that. Working exclusively from the bullpen in 2013, Collmenter pitched 92 innings over 49 games. He had 28 outings categorized by The Bill James Handbook 2014 as Long Outings, more than any other reliever in baseball. Nor was he, as most long men these days are, a mop-up man: He had a 1.3 average Leverage Index for the season, making him the only pitcher with 20 or more Long Outings and an LI north of 1.0. (Stammen had 21 Long Outings and exactly a 1.0 LI.)

Collmenter also entered in plenty of jams, inheriting 22 baserunners on the season. Only two of them scored. Among relievers who entered with at least 20 runners on base, only Sergio Santos, David Aardsma, Pedro Feliciano and Sean Doolittle stranded a higher percentage of those would-be tallies. Those four pitchers combined for six outings the Handbook categorized as long ones, five of those by Aardsma. Collmenter was the only pitcher in baseball who reliably came into tough spots, put out the fire, and then kept the fire out beyond his first trip into the dugout.

He did it without the typical, large bump in pure stuff that permits many untenable starters to become serviceable relievers. He averaged only 87.5 miles per hour with his fastball, yet he threw it over 70 percent of the time. The rest of his pitches were almost exclusively changeups. With two uninspiring pitches, and without a singular skill like an elite ground-ball rate (indeed, Collmenter is a fly-ball pitcher) or top-shelf command (though he did issue just 27 free passes on unintentional walks and hit batsmen over 365 batters faced, indicating good control), Collmenter posted the 25th-highest Win Probability Added of any reliever in baseball in 2013, and that undersells him.

That number doesn’t account for the way Collmenter extended long games, keeping Arizona alive. It doesn’t account for the roster flexibility he provided. It doesn’t account for the lightened pressure on the starting rotation to always get deep into close games, or for the increased comfort Kirk Gibson could feel in pinch-hitting when the situation demanded it during the middle innings of games.

Long relief is a lost art. The transition to one-inning marvels with sliders your grandfather would assume were spitballs has stunted offense, but only by making rosters so creaky that many perfectly competent hitters no longer have a place on them, and by forcing batters to be better at their jobs than pitchers are at theirs. Collmenter is a breath of fresh air, and with a two-year deal with two very affordable options now tied to him, he’s as underrated an asset as any pitcher in baseball.