I got to participate in the voting for last season’s American League Rookie of the Year award. Like almost everyone else, I agonized over whether to go with Gary Sanchez or Michael Fulmer, before ultimately settling on the latter. Tyler Naquin rounded out my ballot. It was the most common ballot — there were a dozen, out of 30, that read like mine.

If I had a fourth-place slot, I would’ve found room for Christopher Devenski. As is, Devenski showed up on five ballots, finishing third on four and second on one. There were many rookies last year who were pretty good. Devenski, somehow, has seemingly flown under the radar. He was so good!

If I wanted, I could leave it at this — there were 142 different pitchers in all of baseball who threw at least 100 innings. Here are the best of them, by ERA-:

1. Clayton Kershaw, 43
2. Kyle Hendricks, 51
3. Chris Devenski, 52

But you could look at the fact that Devenski allowed just four homers and conclude he got lucky. He probably did. Doesn’t mean he didn’t get better and better. It’s Devenski’s second half that really stands out. After the All-Star break, he threw 49.2 innings. Here are some percentile rankings out of everyone who cleared 25 post-ASB frames.

Down the stretch, Devenski threw strikes like Max Scherzer, and he missed bats like…well, also Max Scherzer. He compelled hitters to constantly chase out of the zone, in part a testament to his outstanding changeup. But you shouldn’t come away believing Devenski was some kind of soft-tosser; by September he was throwing his average fastball around 94 miles per hour.

Devenski was underrated because he wasn’t really a starting pitcher. He did start a few times, but he predominantly relieved, without closing. As you know, statistically speaking, relievers have it easier. Yet Devenski wasn’t a reliever of the conventional sort. The last bar in the plot above reads PA/GR. That’s out of a different sample pool — all relievers with at least 10 appearances in the second half. The category itself refers to average plate appearances per relief appearance. Devenski was close to the top. He wasn’t a matchup guy, or a one-inning guy. He was more of a multi-inning swingman, which makes his numbers all the more remarkable. He was that good, for dozens of pitches at a time.

Borrowing from Texas Leaguers, Devenski made a midseason change that could’ve driven his greater success.

As the season wore on, Devenski all but abandoned his curveball, replacing it with a sharper slider. He showed excellent command of the pitch, and opponents missed it nearly half of the time they swung. Here’s Devenski using the slider to get back in the count:

The slider gave Devenski a consistent third pitch. In the second half, he threw it 22% of the time. He threw his fastball 40% of the time. And he threw his changeup — this changeup — 32% of the time.

Lefties would see all three pitches. Righties would see all three pitches. And Devenski’s over-the-top delivery presumably added some measure of deception. His fastball comes with a good amount of rise, and the changeup plays off of it perfectly. The slider is why the Astros are still thinking about Devenski as potentially becoming a starter. He might work well for five or six innings at a time. It’s already clear he can work well for two or three.

Devenski is going to be a key part of the Astros’ enviable pitching depth, no matter his role. Whether he’s rotation insurance or a steady presence in the bullpen, he’ll provide something every manager wants. As he showed over last season’s final three months, he can pitch as well as almost anyone else, and this from someone who was a completely unheralded prospect. A 25th-round draft pick who was included as a PTBNL for Brett Myers. You can give the White Sox some credit for identifying and developing Jose Quintana. In a similar but opposite way, I guess Devenski was an oversight.