When the Braves signed Drew Smyly in November, it’s safe to say that my reaction was on the lower end of the satisfaction spectrum. It’s difficult to deny how good Smyly was in 2020, but it’s similarly hard to forget how profoundly awful he was in over four times as many innings in 2019, or the fact that he had thrown literally one professional inning across the two years before that. From my perspective, the Braves seemingly paid extra for the privilege of finding out whether he’d look more like the 2020 version of himself rather than the pre-2020 version, and that’s definitely not a thing I’d personally be interested in paying extra for. In any case, though, it’s already happened, and now that it has, I’m happy to announce that we all have something wonderful to look forward to: Drew Smyly’s bizarre, backwards curveball.
To be clear, you probably should have already been excited about Smyly’s curveball in a vacuum. Smyly allowed just a .205 xwOBA on his curveball in 2020; only five pitchers with as many or more curves thrown as Smyly had a better mark. His whiff rate on the pitch? A hilarious 50 percent. In general, it was around a top-10 curve in the short 2020 season in everything but outcome, where it was still good-but-not-great due to some unfortunate batted ball results. Of his three offerings, it was far and away the best, and really the only reason he was as successful as he was. His xwOBAs allowed on his other two pitches, a four-seamer and a cutter, were unimpressive at .360 and .330, respectively. All of this is hardly a secret — Alex Anthopoulos went on the radio and specifically singled out his curveball as the thing as far as Smyly was concerned:
Alex Anthopoulos on the signing of Drew Smyly: "We looked at him even last offseason as well. When he came back from the finger injury, his last four outings were really strong. The curveball is a real weapon for him. In our minds its one of the better curveballs in the game"— 680 The Fan (@680TheFan) November 16, 2020
Now, a good curveball is certainly an asset. It’s not like this is a new development, either: Smyly was horribad in 2019 (131 ERA-, 134 FIP-, 115 xFIP-), yet his curveball was running a whiff rate close to 40 percent and a sub-.300 xwOBA then too. (He was horrible because his four-seamer and cutter were just destroyed left and right, inning after inning.) But a good curveball, well, that perhaps just warrants a post that says, “Hey, Drew Smyly’s curveball is really good!” This is not that post. Instead, this is a post about the fact that his curveball is somehow backward.
Pitchers and their arsenals are all unique snowflakes, except they really aren’t. If they were, we couldn’t have this generic pitch classification chart.
Granted, those ovals are all fairly wide, and allow for things like a slider’s motion being literally whatever you want it to be (it’s got parts in all four quadrants!), but a “curveball” is supposed to break away from a same-handed hitter. The most generic curveball from a lefty in baseball belongs to one Keegan Akin, whom you might remember as that guy on the Orioles that threw five scoreless against the Braves last year. Here’s what Akin’s curveball looks like when plotted on a movement chart:
And here’s what it looked like in game action:
The video above wasn’t chosen just because it came against the Braves — that was the most generic curve Akin threw all season, in terms of the lowest deviation from his average curve across velocity, horizontal movement, and vertical movement. Akin’s curve, too, is painfully average: it is 0.1 standard deviations slower than the mean curve from a southpaw, and less than 0.1 standard deviations different in terms of both horizontal and vertical movement. As you can see, even with the off-center camera angle, Akin’s generic curve does what you’d expect it to from the plots: it moves both sideways to the glove side and downward in an exaggerated fashion. So, keep that in mind as you peruse the below.
Meanwhile, here’s Drew Smyly:
I’ll admit, the first time I saw this, I thought it was a mistake. To be very clear, what this plot is suggesting is not that Smyly’s curve has horizontal movement so slight as to essentially be nil. Rather, it has crossed that threshold and broken on through the other side, somehow moving to his glove hand. Honestly, though, it doesn’t look very dramatic in real time. If you were expecting some Wile E. Coyote-esque cartoon physics, sorry to disappoint. It’s more in the “slightly uncomfortable, uncanny valley” category than one of the Bugs Bunny baseball shorts:
I reference the uncanny valley because, when you watch Akin’s generic-o-curve above, you can see it fritter towards Swanson as it hurtles out of Akin’s hand. With Smyly’s curve, though, there’s no jerky trajectory or anything. It comes out of the hand and seems like it’s going to sweep, and then it just doesn’t.
Not that this means we can’t dig for some weird Smyly curves. Again, nothing cartoonishly dramatic here, but this clip made me think my eyes weren’t working right at least once:
This one fooled Greg Garcia without its lack of horizontal motion, but still, it’s kind of a minor “whuh?” moment:
In 2020, lefties threw 7,529 curves as classified by Baseball Savant. Of those curves, only 294 (right around four percent) went “backwards,” i.e., towards the first-base side. Of those 294, 173 (just below 60 percent) were thrown by Drew Smyly, who essentially never threw a single curve that went towards the third-base side. (Different queries yield different results; the number is either zero or two, based on rounding.) Of the curves with the most arm-side motion, Smyly had 22 of the top 23, and 109 of the top 111. However, he was not able to claim the top spot, due to this pitch from Brooks Raley, which I think might just be a data error — I legitimately can’t tell:
Basically, in the backwards curves from lefties realm, it’s pretty much just drew Smyly, and this one pitch from Robbie Ray that I think does a good job of illustrating how slightly off-kilter a backwards curve is:
Part of what makes it hard to show the backwards-ness of it all is that center field cameras aren’t straightaway, so they warp the perspective a bit. Unfortunately, Smyly didn’t have a chance to pitch at the parks with the best-aligned cameras in this regard (PNC Park, The Trop, Miller Park, Minute Maid Park, and of course, Truist (lol) — which may explain why you might feel broadcasts for road Braves games feel slightly off), but the data are what they are. (Except for that one Raley pitch, still no idea what’s going on there.)
As a side note, backwards curves from righties are just about as rare as they are from lefties, but they’re harder to discuss. Why? Because most backwards curves from righties are actually just the result of position players lobbing a ball up there — like Jedd Gyorko throwing something that is definitely not a curveball but flummoxes the algorithm into a misclassification:
If you loosen the algorithm a bit and allow knuckle curves and “slow curves” to factor in, backwards curves from lefties are more common, but only slightly: the rate increases from four percent to around six percent. Smyly’s dominance here slinks backwards, falling from 60 percent of these pitches to 30 percent. Still, he’s got 22 of the top 30 most-backwards curves, and 83 of the top 100. I mostly wanted to add in the other curve types because doing so highlights another old friend, Alex Wood. By some measures, Wood may have the most backwards curve of them all, as you can see from the chart here and the snip below.
I don’t know nearly enough about the different Baseball Savant and Pitch F/X metrics to understand exactly why the movement here differs from the movement values in the raw pitch logs — it probably has to do with cutoffs about which pitches are/are not used, but I can’t be sure. In any case, Wood may have the most backwards lefty curveball when adjusting for relative velocity, and therefore gravity, but Smyly’s volume of backwards curves is much more whelming. Derek Holland also exists. Maybe the Braves will sign both of those guys for depth, and corner the market on bizarre left-handed curveballs. Before they do that, though, they may want to do some research into whether the backward-ness of the pitch is actually related to its effectiveness — I suspect the answer is that the two are not very related, but that’s a topic for another day.
In any case, the point is: I for one welcome our new, backwards-curve overlord. It may not look too weird, but it really kind of is. 2021, provided it happens, should be a fun season for the Braves (though expanded playoffs may seriously crimp that, provided they happen, which hopefully they don’t), and Drew Smyly’s weirdo curve is just another form of icing to be added to the cake.