Which Way Is Janson Junk Going to Regress?

I don’t know why, but I’ve spent more time thinking about the Miami Marlins than any other team this winter. It started when the Marlins came to the end of the season with a not-terrible 79-83 record — three games better than the Braves! — with some interesting pieces looking forward.

Then two Marlins showed up on my search for the next Geraldo Perdomo. Then I kind of talked myself into Miami being one lucky free agent signing from competitiveness — just like in 2002. But it turns out they’re not even making a token effort to spend. Then they traded from their surplus of pitchers to beef up their anemic lineup, which seemed like an OK compromise to make at the time even though it thinned out their greatest strength.

But then the Marlins thinned that rotation out even further, shipping Ryan Weathers to the Yankees for four prospects. So now the Marlins are in a position where Janson Junk is back in their presumptive Opening Day rotation. How fortuitous, because I’d been meaning to write about Junk for a while.

Like I said, it’s been a weird year.

Junk himself is a bit of an oddity; players like him provoke irresistible curiosity in the kind of people who sign up to write for FanGraphs. Specifically, Junk just straight up didn’t walk anyone this year; he ended the season with just 13 walks in 110 innings. His 2.9% walk rate was the lowest among pitchers who threw 100 or more innings in 2025, and the third-lowest in any 100-inning season in the 2020s, trailing only George Kirby in 2023 and Bryan Woo in 2024.

There must be something in the water in Seattle; Junk played his college ball at Seattle University, from whence he was a 22nd-round pick in 2017. That’s the kind of small school where Junk would ordinarily go down as a legend for just making the majors; in fact, he’s one of just five Redhawks to ever reach the Show.

Unfortunately, one of Junk’s college teammates was Tarik Skubal, who’s overshadowed him a bit.

The guy who never walks anyone is a great hook, so David Laurila spoke with the Marlins right-hander in August. He talked about the new grips he’s been using on his changeup and sweeper, and how he’d refined his command by training at Driveline (shout out Seattle again) over the winter.

At the end of the season, Michael Rosen dove into the discrepancy between Junk’s position atop the walk rate leaderboard and his appearance near the bottom of the chart in Michael’s bespoke command metric, the Kirby Index. He discovered that Junk’s arm angle was all over the place, but bizarrely, this had no demonstrable effect on the vertical movement of Junk’s fastball. I’m not going to get quite that granular, because the headline number that interests me about Junk is his FIP.

In 2025, Junk posted a 3.14 FIP, which was 30th out of 199 pitchers with 70 or more innings. That’s good for anyone; the guy one spot ahead of Junk on the FIP leaderboard was Hunter Brown, who finished third in AL Cy Young voting. For the Marlins’ fifth starter, that’s exceptional.

As you know, FIP has three inputs: home runs, strikeouts, and walks plus hit batters. Junk had the lowest walk rate of any regular starter and he only hit four batters all season. But his strikeout rate, 17.2%, was 170th out of 199 among pitchers with 70 or more innings. That’s quite a pair of extremes. What about the third input, home runs?

Junk did well here too. He allowed only eight dingers in 2025, for a HR/9 ratio of 0.65. That’s good for 27th in the league, with that 70-inning minimum. But that actually understates how good Junk was at keeping the ball in the yard. A lot of the guys ahead of him on that list were either relievers or groundball gluttons whose fastballs come off hitters’ bats and immediately bore into the Earth in search of oil.

Junk isn’t that guy. In his discussion of Junk’s arm angle, Michael invoked the name of Trey Yesavage, whose combination of high release point and nuclear splitter turned him into a celebrity overnight this past October.

Now, a guy with a great splitter ought to get a lot of groundballs, right? Yesavage did in his brief stint in the majors, but in 25 minor league appearances, he actually had a slight fly ball bias: 0.87 grounders for every fly ball. It’s not just the splitter, it’s the four-seamer with rise and cut that hitters can get under.

Junk doesn’t throw quite as hard, and throws three breaking balls to Yesavage’s one, but he has a lot of the same characteristics: more rise and cut than average on the four-seamer, and a GB/FB ratio that’s within spitting range of even.

If you look at HR/FB%, Junk was 11th in the league with a 70-inning minimum and third out of 127 with a 100-inning minimum. Only 6.0% of his fly balls went out of the park.

HR/FB% is kind of like BABIP, in that about 15 years ago, the conventional wisdom was that it’s random. As our understanding of pitching has improved — especially our ability to measure quality of contact — that’s not quite the case anymore. Still, the null hypothesis is that a huge variation from the norm in HR/FB% (the league average was 11.9% last year) is probably a fluke, and ripe for regression.

Is there something Junk does well in terms of reducing quality of contact on fly balls?

Junk, uh, lets up a lot of contact, and especially a lot of hard contact:

Junk’s ERA in 2025 was 4.17, which was 1.03 runs higher than his FIP. That’s the 17th-largest underperformance of FIP in the league (again, minimum 70 innings). But a lot of pitchers with bigger gaps just flat-out stunk and had huge ERAs. If you look at FIP as a percentage of ERA, Junk was in the bottom 10; his FIP was 75.4% of his ERA.

Now, xERA, which measures quality of contact, was far less sympathetic. By that measure, Junk actually overperformed his ERA by about 6%. Of the 18 pitchers with the lowest FIP/ERA ratios last year, Junk is the only one with an xERA higher than his ERA.

That above table is why.

And yet, his xERA was 4.45. A pitcher with a top-10 contact rate and a top-10 EV90 ought to have an xERA you can only count using two hands. The only qualified hitter in the top 10 in EV90 last year who was even in the top 50 in contact rate was Vladimir Guerrero Jr.

So why wasn’t Junk making opposing hitters look like Vladito?

Two reasons. First, apart from not walking anyone, Junk did one thing very well in 2025: He got batters to chase. Did he get batters to swing and miss at pitches outside the zone? Oh, no, of course not. Junk’s out-of-zone whiff rate was worse than 20 different pitchers’ in-zone whiff rate.

But even if you hit the ball, it’s hard to do a lot of damage on pitches outside the zone. The league-wide wOBA on batted balls outside the strike zone was .285; inside the zone, it was .381. A healthy 19.8% of the balls put in play against Junk came on pitches outside the strike zone, which was in the top third of the league. And here I think he was getting unlucky; Junk allowed the highest wOBA in baseball (minimum 300 total balls in play) on pitches outside the zone, outstripping his xwOBA by 80 points.

The second thing that went well for Junk was that he used the whole ballpark.

LoanDepot Park is a pitcher-friendly stadium. Using Baseball Savant’s xHR, only Kauffman Stadium and Busch Stadium’s dimensions would’ve led to fewer home runs than Junk’s real-world home park. If he’d given up the same batted balls at Citizens Bank Park, Junk would’ve allowed 17 home runs, not eight:

That’s a massive red flag for regression.

I think Cabrera is far from the ace-in-waiting that certain individual pitches make him look like. I’m not a huge Weathers fan either. And whatever else you want to say about Junk, it’s a developmental triumph that the Marlins even got one season like 2025 out of a guy who took until age 29 to establish himself as a major leaguer. (Junk turned 30 on Thursday as I was writing this. Many happy returns.)

But the Marlins have now shed enough rotation depth that they’re into the Quad-A guys and prospects. Junk’s command and Yesavage Lite characteristics make him a pitcher worth monitoring, but given the choice between believing in his FIP and believing in his xERA, I have to choose the less flattering of the two.