Pitches in Radar Gun Are Slower Than They Appear: Identifying Baseball’s Perceived Velocity KingsJim McIsaac/Getty Images
When Pirates rookie Jung-ho Kang, a veteran of the Korean Baseball Organization, was asked earlier this month to name the nastiest pitcher he’d seen since joining Pittsburgh this spring, he didn’t pick someone predictable, like starters Matt Harvey, Cole Hamels, Jon Lester, or Jake Arrieta. Nor did he choose closer Aroldis Chapman, whom he’s faced twice. Instead, Kang gave a cryptic response: “the weird, jump-step guy.”
Kang didn’t know the name of the nastiest pitcher, but he remembered his stuff. And if you know the name of the nastiest pitcher, it’s probably only because of the one weird trick Kang referenced. The nastiest pitcher, according to both Kang and the many weak hacks other hitters have had against Kang’s nemesis, is Marlins right-handed reliever Carter Capps, whose delivery is the strangest thing you’ll see on a major league mound this season, aside from Pat Venditte’s.
Capps, whom the Mariners traded to Miami for Logan Morrison in December 2013, has always tested the limits of major league mounds. When we first saw him, he threw from as close to third base as he could in an attempt to unsettle right-handed hitters. These days, his horizontal release point is no longer extreme. Instead, the current incarnation of Capps is exploring the studio space in a different direction, unveiling a borderline illegal delivery that propels his release point almost unimaginably close to home plate. Shifting rightward on the rubber theoretically made Capps more vulnerable to opposite-handed hitters. Shifting forward theoretically makes him more effective against everyone.
Measured by conventional means, Capps has a high-90s fastball, one of baseball’s hardest. But the radar gun still says the righty throws 2 mph slower than Chapman. In one important respect, however, the radar gun is wrong.
Statcast, the camera-and-radar-based tracking system installed by Major League Baseball Advanced Media in all big league parks for the first time this season, assigns each pitch a “perceived velocity,” which MLB.com defines as the “velocity of the pitch at the release point normalized to the average release point for MLB pitchers.” The intention is to capture not only how fast a pitch actually travels, but how fast it appears to travel. All else being equal, pitches released closer to the plate should look and act faster, as far as batters — who have an even smaller fraction of a second to react when they pick up pitches later — are concerned. As amusing as Capps’s new technique is, he doesn’t hop halfway to home for fun. He does it because he believes it gives him an advantage, which Statcast can quantify.
Thus far, MLBAM has kept Statcast’s results pretty close to the vest, incorporating some of the system’s data into baseball broadcasts and MLB.com clips without fully opening the vault. For this article, though, MLBAM provided a peek, offering average perceived velocity figures on four-seam fastballs for every pitcher whom its pitch-classification algorithm flagged as having thrown a four-seamer through Sunday’s games.
Perceived velocity is largely based on another Statcast measurement, “extension,” defined as “the distance of the release point of the pitch from the front edge of the pitching rubber.” Capps’s extension is 8.2 feet, the longest in baseball by more than half a foot (in second place is Cardinals reliever Jordan Walden, who also has a hop). When Capps is pitching, 60 feet, six inches shrinks to roughly 52 feet, four inches. That’s enough to inflate his perceived velocity by almost 3.5 mph, topping 101, passing Chapman (who gets a much smaller boost), and crowning Capps as baseball’s hardest-seeming thrower.
The difference in extension between Capps and Dodgers reliever Joel Peralta, who lies at the opposite end of the list, is about 3.25 feet, or 39 inches. Compare Capps’s plant point, which is almost at the edge of the dirt, to Peralta’s landing spot, which leaves what looks like acres ahead of him.
Peralta, who has a short stride even when he salsas, has released the ball less than 5 feet from the rubber this season — closer, even, than Mariners catcher Jesus Sucre, who qualified for second-to-last place with a mop-up appearance last week. The difference in perceived speed between Capps and Peralta amounts to more than 6 mph. Given the difference in performance that even 1 mph can make — particularly for relievers, who tend to be more reliant on their fastballs than starters are — it seems clear that perceived velocity is worth closer examination.
Kang, who’s made a smooth adjustment from the KBO to MLB, is 1-for-2 with a single and a hit by pitch against Capps, but it’s not surprising that those plate appearances stood out as particularly uncomfortable. Capps has allowed the lowest contact rate of any pitcher who’s thrown at least 10 innings this season. Batters have whiffed on almost half the swings they’ve taken against him, managing only a 51.8 percent contact rate. Chapman’s 56.4 percent contact rate from last season is the lowest any pitcher has allowed in a season of at least 10 innings pitched since 2002, so given the trend toward shorter relief outings and decreasing contact rates, it’s not a stretch to say that “the weird, jump-step guy” might be the most unhittable hurler ever. Not only do his opponents have to contend with the shock value of watching a pitcher grab air, they also have to gear up for the fastest perceived pitch in baseball while worrying about something off-speed: Capps’s curveball has an insane 74.4 percent whiff/swing rate, much higher than the next-closest curve that’s been thrown at least 50 times.
Capps doesn’t need to be weird in the way that the soft-tossing Venditte does: He stands 6-foot-5, throws hard, and has a filthy breaking ball coupled with great control. Pushing up his release point to the degree that he does is almost unfair, like Andrew Friedman having baseball’s highest payroll. Capps is the predictable star of extension, but you can explore the full sample (for four-seamers only) in the embedded table below. From the glimpses I’ve gotten, each pitcher’s extension tends to be fairly stable across multiple pitch types. One somewhat counterintuitive observation: The correlation between extension and height is only 0.25, a fairly weak relationship. Variations in mechanics explain more of extension than do variations in height, so short pitchers aren’t necessarily sentenced to low perceived velocities. The list of sub-6-footers with the highest ratios of extension to height includes famous long-striders like David Robertson, Tim Lincecum, and Craig Kimbrel, as well like lower-profile pitchers like Brandon Kintzler, Danny Farquhar, and David Goforth.
As is often the case with new technology and advanced stats in sports, Statcast in many cases only confirms truths that players and executives have always known. Hitters often refer to pitchers whose stuff seems to transcend its velocity as “sneaky fast,” or complain that their offerings seem to arrive early. This can mean many things: In some cases, it describes pitchers who hide the ball with their body, or who disguise a big fastball with a smooth, low-effort release. In others, it might hint at superior extension. For instance, it makes perfect sense that San Francisco’s Yusmeiro Petit would appear close to the top of the list of biggest differences between perceived and actual velocities. Petit has long been celebrated for his deception, and while his release point probably isn’t his only secret, it helps explain why he’s enjoyed better results than the typical righty with a fastball that sits around 89. The Giants swingman is only 6-foot-1, which makes him an anomaly at the leaderboard’s upper bounds: Among players who rank higher than 40th place, Kyuji Fujikawa is the only other pitcher who is Petit’s height or shorter. The top of the list includes several other pitchers who are often credited as deceptive (Jered Weaver, Michael Wacha, Tony Cingrani, Doug Fister, Robertson) as well as some guys with great stuff who’ve hardly needed the help (Stephen Strasburg, Andrew Miller, Chapman).
For an example of a pitcher whose perceived velocity might explain a lack of success, we can turn to Royals starter Yordano Ventura. Ventura’s velocity has declined sharply this season, but it remains among the elite. His average four-seamer speed ranks sixth among starters, but his four-seamer whiff rate ranks 95th, closer to the bottom than the top, which explains why his strikeout rate is somewhat underwhelming. Statcast reveals that among full-time starters, only Tom Koehler1 has a greater negative difference between actual and perceived velocities than Ventura, whose four-seamers appear 2 mph slower than they actually are. Maybe Ventura, whose delivery is commensurate with his smallish frame, doesn’t miss bats like a guy who hits the high 90s because batters don’t see his heat as high 90s.
Like many men with small extensions, last-place Peralta compensates for his shortcomings by driving a sports car, and also by varying his arm slot and taking his sweet time between pitches, another way of disrupting his opponents’ timing. He’s had effective seasons in which he’s missed many bats. Extension isn’t everything. It’s possible, also, that the sore shoulder and neck that have sidelined Peralta for much of the season have reduced his extension relative to what it was in the past. In time, we can determine whether changes in extension could be used as a crude tool to identify certain types of injuries.
So how much does all of this matter? Actually, less than one might think. According to one team executive who’s studied extension, the information is “probably of not much use given [actual] velo.” Although he acknowledged that the numbers might have value in some cases, he told me “it’s been difficult to tease out.” With help from a few of the Internet’s best baseball number-crunchers, I searched for relationships between perceived velocity and performance and found nothing to contradict what the team exec told me. Perceived velocity explains only slightly more of the variance in stats such as whiff rate and ground ball rate than actual velocity does, and neither explains much on its own (or even in tandem with pitch movement), probably because many skills that play important roles in pitch speed — like command, deception, sequencing, and spin — are currently difficult to measure.
After looking at Capps’s stats,2 it might be hard to believe that perceived velocity isn’t baseball’s next big idea, but Capps is one of a kind. His command doesn’t seem to have suffered since he took flight, but it’s possible that for many pitchers, an attempt to increase extension would lead to an elevated walk rate, which has always been an issue for Walden. More importantly, for most players, the radar matches the drapes. About 40 percent of pitchers in this sample have perceived velocities within 0.5 mph of their actual velocities. And among all pitchers, the correlation between actual and perceived four-seamer velocity is 0.95, which means that pitchers who rate highly in one category usually rate highly in the other.
In the majority of cases, perceived velocity simply doesn’t tell a dramatically different story from the one we know. As a result, it might make a major difference only for the small set of edge cases that significantly deviate from the norm, like Capps, Petit, and Ventura. And it doesn’t take a sophisticated tracking system to prove that Capps and Walden are doing things differently.
None of this means we should declare perceived velocity a dud. Statcast is exciting, but we shouldn’t expect it to radically and routinely realign our understanding of what it means to be good at baseball. In many cases, it will confirm what we know, in smaller samples than were meaningful with blunter measurements, and with a few surprises mixed in. Maybe Capps will be a trailblazer, an object of curiosity whose success inspires so many copycats that one day, no-hop windups will look as archaic as Mordecai Brown’s. Maybe Capps is only an outlier whose uniqueness is now slightly easier to appreciate. Still, even if all that perceived velocity does is help us describe the occasional Capps and help a team unearth the next Petit, it will be a worthwhile addition to our tool set. To borrow a phrase from Carl Sagan, it does no harm to the romance of a weird, jump-step delivery to know a little bit about it.