James Paxton’s Impossi-ball
James Paxton was a pleasant surprise at the end of last season, and the more you dig into his brief MLB career, the stranger it looks. I don’t mean to imply that Paxton got by on luck, or that he doesn’t have the talent to be an excellent pitcher in the big leagues. It’s just that he pitched so differently from the views we’d seen in the minors, the AFL and in spring training. The obvious one – and the factor that had the biggest impact on his runs allowed – was his strand rate. The guy who was completely undone by big innings early in the PCL season stranded nearly every baserunner he allowed (small sample size, of course). The guy who stranded just shy of 66% of runners in the minors shot to the majors and stranded 88.5%! It’s odd, but it’s not the strangest thing, at least to me. The strangest thing was that his ground ball rate was higher than Justin Masterson’s.
Paxton had essentially average GB rates coming up, hitting 50% for Tacoma by Statcorner, or 46% by MinorleagueCentral.com. The exact number doesn’t really matter; the point is, he was typically in the high 40% range, against league averages from 46-44% (as you probably know, the league-wide ground ball rate slowly drops as you rise through the minor league levels). In the majors, he posted a 59% rate, higher than any month he had in AAA. I get it – he made 4 (four) starts. Maybe it’s a fluke, but the M’s clearly don’t think so, and they (or rather Rick Waits) made some sort of an adjustment to his motion mid-way through the 2013 campaign. If he’s really going to run GB rates a standard deviation or two above the mean, that’d be huge for his projections.
It would also be bizarre, given his throwing motion. There are a few attributes of a pitch that lead to higher ground ball rates. The first is location. If a pitch is down in the zone (or, even better, out of the zone), it’s more likely to be a ground ball. As Dave Allen wrote years ago, a pitch’s whiff rate increases and the ground ball rate *decreases* the higher it is. But this effect may be dwarfed by a pitch’s movement. A sinker or two-seam fastball can often get strikeouts in the middle of the zone simply because the pitch appears to sink more than most fastballs, and batters don’t always adjust. As Dan Lependorf found, it often helps a fastball to have a lot of horizontal movement as well, as two-seamers and change-ups do. This is why Carson Smith has always run, and WILL always run crazy ground ball rates. His sinker is tailor-made to get ground balls. But Paxton’s fastball – and his whole throwing motion – is the opposite of that. He has an over-the-top delivery and has shown a lot of vertical movement. Paxton got tons of grounders on his fastball, and he got them pretty much wherever he threw them. The answer pretty much has to be in the movement on his fastball.
When Paxton hit the Arizona Fall League in 2012, we finally got to see some legitimate velocity and movement readings on him. The results were encouraging, and matched up with the his mechanics pretty well. His fastball checked in around 95mph (woohoo!), showed very little horizontal movement, and a ton of “rise” or horizontal movement. This made sense: the more a pitcher comes “over the top,” he imparts less side spin, and much more backspin . Look at a list of the guys with the highest GB rates and you see plenty of side-armers for a reason. Guys with over-the-top deliveries like Josh Collmenter, Chris Tillman, or Stephen Pryor don’t get ground balls. So how’s this possible?
Well, we can just check pitch fx and see what that change he made in Tacoma was all about, right? Well, this is where the park-by-park calibration of the pitch fx system comes in. Paxton hasn’t thrown too many pitches in front of pitch fx, and unfortunately for us, the two parks we’ve seen him at most – Safeco and Peoria Stadium – are two of the biggest oddballs around. Safeco’s raw movement readings look strange -they’re consistently shifted down vertically by around 3″. Brooks Baseball adjusts for this, but there are a few games when the system’s calibration error is much larger than normal. As it happens, Paxton’s MLB debut was one of them. The raw data show Paxton’s fastball in the first inning had about -5 inches of vertical movement, or about the same as the average curve ball. This is just garbled stuff, really – instead of shifting the vertical movement down 3″, it was down around a foot or so.* The data were clearly adjusted by the time they were posted on Fangraphs/Brooks, but they’re not terribly comparable with any other movement readings we’ve seen from Paxton. Since he only had 4 big league starts, this game exerts a pretty big impact on his season averages. So, to sum up, Paxton’s averages in the spring/AFL and his averages in his brief call-up look nothing alike. Both Peoria and the MLB data is somewhat problematic. Awesome.
OK, I’ve got a few guesses as to what’s going on.
1) Paxton used to throw a four-seam fastball, but either shifted to a two-seamer or made a slight tweak to his release in July/August of last year. That’s what explains the slight bump in horizontal movement. Sure, it doesn’t have actual sinking movement, but maybe much of the contact on pitches coded as “four-seamers” are actually stealth sinkers. Hard-throwing reliever Jeremy Jeffress of Toronto used to have movement very much like Paxton’s, with almost zero horizontal movement and a ton of “rise” on his plus fastball, but like Paxton, he got above-average GB rates all the way up the ladder. Last year, he also made a slight tweak to his fastball, adding a bit more horizontal movement, and he posted a nearly 70% GB rate (in a small sample, of course). This hints at the third attribute of a grounder-inducing pitch: velocity. The harder it’s thrown, the more grounders it’d get, all other things equal. But the problem is that the baseline – given the vertical movement we’re talking about – is so low, that velo alone doesn’t seem enough to explain it. Again, it’s not like Stephen Pryor has elite (or even average) GB rates.
2) Paxton’s GB rate, which he’s maintained throughout the spring, is going to fall as he moves through the regular season. For reasons I’m not sure on, hitters may hit the ball on the ground when facing new pitchers, especially if they’re throwing hard. Yordano Ventura of the Royals saw a bump in his GB rate when he hit the majors as well, and he’s got Paxtonian vertical movement on his 100mph fastball. That’s probably not going to stick around either, as he always had slightly below average GB rates in the minors. But there may be something to the combination of hitters seeing someone for the first time and velocity that makes it really hard to elevate the ball. This is the “it’s all a small-sample mirage” explanation; looking at his minor-league career numbers, it’s tempting to go with this. But it doesn’t explain why his career numbers are still higher than you’d guess, and it doesn’t explain why he’s continuing to rack up worm-burners in the spring.
3) There’s something in Paxton’s delivery that gives hitters some trouble in locating the ball. The other “anomaly” in the over-the-top pitchers is the guy the M’s told Paxton to model his delivery after – Clayton Kershaw of the Dodgers. Kershaw’s fastball looks like Tillman’s via pitch fx. It’s no longer got plus velocity, it’s arrow straight and it has plenty of “rise.” But while Kershaw’s GB% is below 50%, it’s a bit above average, and he’s able to get GBs from everywhere in the zone on his four-seam fastball. It’s not a sequencing issue, where hitters expecting FBs hit over the top of sliders – the beat a rising fastball into the ground. Gabe Kapler talks about Kershaw’s delivery here. That said, the aspect Kapler mentions – Kershaw’s mid-motion pause – isn’t something I associate with Paxton. On the other hand, Kershaw’s GB rate was higher with men on base/RISP last year, when he’s pitching from the stretch and not his stop-start wind-up. It may be that Kershaw, Paxton and Jeffress are able to hide the ball longer than other pitchers with similar motions, and this is what gives the pitch the appearance of sink – the dreaded “heavy ball.”
What about the old stand-by, the vaunted “Downhill plane?” Well, the problem is that hitters’ swings have loft, which counteracts the benefit of a pitch coming in at an angle. And again, if “downhill plane” on its own made for a flurry of ground balls, the purely over-the-top pitchers – the Collmenters and Pryors and Tillmans – would be near the top of the league in GB rates, not clustered together at the bottom. Paxton’s intriguing for a number of reasons, but I’m really curious about his GB rates for 2014. If I’m honest, this is less about how plus-plus GB rates would affect his projections or the M’s wild card chances and more about wanting to learn something new about baseball and how pitching works.
* This happens every now and again. For another example, look at Felix’s 9/1/2012 start against the Angels.
Interesting read, Marc. I’ll be curious to follow this through 2014.
Definitely interesting… Makes me want to compare some early 2013 starts to later starts, in slow motion overlay. I saw him pitch twice in Tacoma last year, early on, and he really looked quite ordinary (sadly)… Then when he came up to Seattle, he looked electric again.