After four years of bouncing between the bullpen and the rotation, David Phelps spent almost all of last year in the Miami Marlins' bullpen, and took to the role like a fish takes to water (yes, I went there. Deal with it).

Phelps set career-best marks last year in ERA (2.28), FIP (2.80) and WAR (1.9 fWAR, per FanGraphs) over 86 2/3 innings, 62 1/3 of which came in relief. He was a top 25 reliever by both ERA- (56) and fWAR (1.3; his other 0.6 Wins Above Replacement came as a starter)

Like most pitchers, Phelps has been more effective in relief than as a starter. For his career, he has a 3.30 ERA, 3.64 FIP and a .286 wOBA against in 147 1/3 innings out of the bullpen. In 350 2/3 innings as a starter, these numbers jump to a 4.21 ERA, 4.04 FIP and .320 wOBA against.

Miami reportedly plans to keep Phelps in the bullpen, with manager Don Mattingly telling MLB.com "We feel like he can affect more games out of the 'pen. ... He gives us -- I don't want to say poor man's -- but an Andrew Miller-type [reliever].”

Things seem settled on the Phelps front, but there is at least an argument to be made Phelps should actually move back to the starting rotation, considering Miami’s need for rotation depth and the value of an average starting pitcher.

Let's look at both sides.

The Case For Starting

Last season, Miami’s starters posted a 105 ERA- and 103 FIP-, and they ranked 21st with a 4.53 Deserved Run Average, per Baseball Prospectus. (For those unfamiliar, DRA measures how many runs a pitcher “should have” given up by adjusting for numerous contextual factors, and given its predictive and prescriptive power, it is arguably the best pitching statistic currently available.)

The Marlins must also deal with the on-field impact of Jose Fernandez's passing, which created a massive hole in the rotation. FanGraphs currently projects Marlins starters to combine to post 9.5 Wins Above Replacement, fewer than all but three teams.

Wei-Yin Chen (2.5 WAR) and Edinson Volquez (2.0) are the only Miami pitchers projected be worth more than 1.3 wins, so the club could certainly use another quality starter.

Since debuting in 2012 with the Yankees, Phelps has amassed 4.0 fWAR as a starter, which comes out to about 1.7 fWAR per 170 innings or 2.3 fWAR per 200 innings. This is in the neighborhood of average value for a starter, but it is also more than most relievers are worth.

Last season, only 17 relievers were worth at least 1.7 fWAR while only seven cleared the 2.3-win mark (and keep in mind, FanGraphs’ WAR model does give relievers some credit for pitching in higher-leverage situations, as does Baseball-Reference's). Eating innings, even at an average level, has value.

There is also a case to be made that Phelps can pitch at an above-average level as a starter this year, one outlined well by Alex Chamberlain at FanGraphs.

During his first four years as a big leaguer, Phelps’ average fastball velocity sat around 91 miles per hour (MPH), before jumping up to 94.63 MPH last season, according to Brooks Baseball. It was easy to chalk this up to his new role, which didn't require him to pace himself.

As Chamberlain writes, however, he maintained this velocity during a five-start stint in August, when he pitched to a 2.22 ERA and held opponents to a .184/.276/.287 slash line -- this was good for a .256 wOBA, identical to what he allowed as a reliever.

A .264 BABIP during that five-start stretch surely helped, but he also walked fewer betters as a starter (10.2%) than as a reliever (11.0%), and he struck hitters out at a slightly higher rate (32.7% to 32.3%).

Phelps may not be able to sustain these numbers for an entire season, but combined with Miami’s need for a starter, they do make a return to the rotation a tempting option.

The Case For Relieving

There is more to this than simply saying “if it’s not broken, don’t fix it,” but this is an admittedly compelling argument (see: Joba Chamberlain). Phelps was never more effective as an MLB pitcher than he was last year, and he could be the latest middling starter to make the switch to relief ace.

Given how Miami projects as a roughly average team, having a good bullpen may also be more beneficial to the Marlins than other teams. FanGraphs’ projections currently have the Marlins finishing 80-82, but in recent years, we have seen squads like the Kansas City Royals, New York Yankees and Baltimore Orioles outperform their underlying metrics by thriving in close games -- thanks in part to a strong bullpen (luck/randomness probably played a large role, as well).

Last year’s Yankees team, for example, had a 79-83 Pythagorean Record, but won 84 games by going 24-12 in one-run games. Their bullpen, which was fifth in baseball in Win Probability Added (WPA), probably had a big say in that.

Miami’s relievers pitched the eighth-most innings in baseball in 2016, but they finished 21st in WPA. Kyle Barraclough (2.1 fWAR) and A.J. Ramos (1.4) also had strong years, but the Marlins' other relievers combined to pitch 360 innings and were worth -1.1 fWAR.

It seems essential for Miami to get the most out of its few effective relievers, so Mattingly would be wise to heed his own words and actually use Phelps like the Cleveland Indians used Miller. When they got him, Cleveland used Miller aggressively in both the regular season and playoffs as he had 10 appearances of two or more innings with the club.

Phelps, despite also being a former starter, only had three such outings in 2016, and he is currently projected to amass 0.7 WAR as a reliever across 55 innings in 2017. Using him more liberally would allow Miami to get more production out of him, turning Phelps into an even more valuable asset.