There was a time when Andrew Cashner had the kind of potential that baseball fans can really dream on. He had a high-nineties fastball, great promise with his changeup and slider, and a high ground ball rate. He even pitched for the Padres, giving him one of baseball’s most pitcher-friendly parks, as if you needed any other reasons to believe in a Cashner breakout. Health was the most significant concern, but if he could just put together a full healthy season, the upside was huge. And if it didn’t work for him as a starter, surely there was a terrific relief arm here.

30-year-old Andrew Cashner is not what anyone projected. The strikeouts are gone. The triple-digit fastballs haven’t been seen for two years. There’s no swing-and-miss any more, with his swinging strike rate at a career-low 5.8%. Only Matt Cain and Bartolo Colon have a lower mark, and both have allowed a ton of runs (Colon’s ERA is the worst among all starters with at least 100 innings pitched). This decline hasn’t been mitigated by great control, as Cashner’s 9.3% walk rate is barely lower than the strikeout rate: his strikeout rate minus walk rate (K-BB%) is 2.6%, the lowest in the league among that same group of 100+ IP starters.

Despite all of this, Cashner has a 3.44 ERA. That’s not in a league where half the starters have a mid-three ERA, as was the case a few years ago, that’s in a league where the average starter ERA is 4.51. Cashner has either stagnated or got worse in almost all of the basic elements we look at for pitchers, and he still has 125 innings of an ERA more than a full run better than the league average.

As we’re now conditioned to look at poor strikeout-walk ratios and predict imminent demise, the natural reaction all year has been to dismiss this as sheer luck. There’s a good chance this is almost entirely luck. However, while lots of strikeouts and few walks is a pretty solid foundation for a successful starting pitcher,  there are other components. As increasingly detailed quality of contact data emerges, it’s becoming clear that contact management is much more of a skill than ERA estimators like FIP suggest. As Cashner isn’t getting the job done by getting hitters to swing and miss, let’s be fair to him and figure out if we should be giving him more credit when it comes to balls in play.

Cashner’s 0.79 HR/9 and 8.4% HR/FB rate go a long way towards explaining how he has allowed so few runs. Those marks rank tenth and third respectively in the same group of starters. Similarly, a .276 BABIP has somewhat mitigated the high walk rate, but that’s hardly remarkable – it’s nowhere near peak Marco Estrada territory. Nonetheless, it implies that Cashner might be doing something to get above-average results on balls in play. The Statcast data gives Cashner a .320 xwOBA, almost identical to his .316 mark. That’s fine. It’s middle of the pack. Jaime Garcia and Alex Cobb have almost identical marks (so do Luis Perdomo and Masahiro Tanaka, so there can be a lot of variation in outcomes).

Similarly, the Statcast-based xStats suggest that Cashner’s results should be about average. His xOBA+ is 100.7, ie. almost exactly league-average. Both numbers point towards Cashner having above-average results on balls in play, but not to the extent that he should be an above-average pitcher. The 30-year-old does have that good ground ball rate and a relatively consistent release point, an important component of pitch tunnelling. Cashner’s cutter (or slider if you really prefer the Statcast designation to the Brooks Baseball one), a new addition to his arsenal last season, hasn’t been particularly effective on a results basis, but it might be giving hitters another look and improving results on his fourseam and sinker. Cashner’s favourite pitch sequence by far after consecutive sinkers is the sinker and cutter, in either order. Where the cutter dives low and away to the glove side, the sinker, which really doesn’t sink all that much, runs to the arm side with some useful late movement, giving him options to pitch to both edges of the plate.

That’s not to say that this is all executed with perfect command. A common outcome if you watch enough outs made off Cashner pitching is the below, in which the catcher seems to be calling for the sinker low and inside to the right-hander, Cashner leaves it up a little, but still draws the poor contact from the hitter who can’t get the barrel round on it in time.

Nonetheless, even if we accept the premise that Cashner is compensating for the strikeouts with a sizeable helping of poor contact, there’s something missing here.The only way Cashner’s numbers could look like they do right now is if the timing of the hits was heavily in his favour. Let’s take a look at his splits with the bases empty and runners on.

With no-one on, Cashner has had awful results. Hitters are putting up a .301/.387/.457 line, which gives him a 130 OPS+ for that split, so he’s 30% worse than league average. That is the 11th-worst split out of the 108 starting pitchers with over 100 innings. Seven of his 11 home runs allowed have come with the bases empty.

It’s probably clear where this is going, but the extent to which there’s a difference is still incredible. With men on, batters are hitting .204/.266/.275, for a 43 OPS+. That’s the fifth-best mark in the league, putting Cashner in the company of a series of pitchers having all-around terrific years: Alex Wood, Gio Gonzalez, Max Scherzer, Clayton Kershaw, Rich Hill, and Chris Sale.

To put this in an even more extreme light: Cashner’s tOPS+, his performance relative to his overall OPS+ allowed, is the worst in the league when the bases are empty, and the best in the league when a runner is on. This is how you walk a ton of hitters, post low strikeout rates, and still have a good ERA without getting a ton of BABIP luck.

So it’s all cluster luck. Someone has to be at the extremes of the distribution, and Cashner just happens to have had his best fortune when runners are on. Or is it? The Statcast data tells another story. Cashner’s xwOBA with bases empty: .349, which is bad, (albeit still better than his .373 actual wOBA). Cashner’s xwOBA with men on is an excellent .285, which is considerably higher than his .244 actual, but also a massive 64 point difference from his bases empty performance. The quality of contact against him has been monumentally different in the two situations, and historically he has performed better than average with men on base, and worse with the bases empty, although not to such an extreme extent.

One clear difference between the two situations is the pitch mix. With the bases empty, heavy sinker usage is the defining feature, with Cashner using that twoseamer 52% of the time against righties in particular, and he goes to it even more when he gets ahead against hitters from both sides. The mix becomes much more balanced with runners in scoring position, with both the changeup and the cutter featuring more prominently, the changeup as much as 30% of the time against lefties, and the cutter at 26% against righties. The fourseam has actually been his best pitch in terms of results, but that goes alongside a significant reduction in usage compared to 2016, currently less than 25%.

What’s more, Cashner throws harder when men are on base. Just using his fourseam numbers to make the comparison fair, Cashner has averaged 93.4 mph with the bases empty, 94.1 mph with a man on. It was the same in the past two years according to Statcast; in fact, the differences are slightly greater. If you’re wondering if that’s a league-wide phenomenon – maybe the pitcher gets amped up with men on, maybe simply a product of more situations where they need a strikeout – there’s a tiny difference, but nowhere near Cashner’s. The league has collectively thrown fourseamers 93.1 mph with the bases empty and 93.3 mph with runners on.

So possibly Cashner should be throwing from the stretch all the time, or maybe adrenaline is simply helping him to pump the velocity up a little. Whatever the answer, it’s a strange aspect of a strange season that probably is helping him at least a little in those runners-on situations.

Watching Cashner doesn’t feel like watching a pitcher with a mid-three ERA, and the sensible bet still seems to be on regression bringing him back much closer to his FIP at least, of 4.58. It’s also fair to say that it hasn’t been all luck so far: batters have made some bad contact against him, particularly when there are runners on base, and the cutter does give him something new to work with. One final note on something else that might be changing about Cashner:

That velocity is coming back on both the fourseam and sinker. Cashner has thrown both up near 98 mph during August and is averaging 94.6 on the fourseam for the month, putting him back in line with where he was in 2016, rather than the disappointing early season numbers. The fact remains that Cashner doesn’t seem to have the stuff to miss bats very often anymore, and this season is likely going to be as good as it gets when it comes to the quality of contact. Still, if Cashner keeps changing, who knows what we might see next. It’s not as though he’s made a habit of doing what we expect.