keyboard_arrow_uptop

Here we are in the middle of the Information Age, with access to more data than the human mind can possibly process, and yet the dissemination of baseball information has been muted by a language barrier. Baseball fans are becoming increasingly savvy about the nuances of the game, with sophisticated analytical tools at their disposal, but access to the dynamics of play on the field is often clouded by a filter of scout-speak. If we were playing poker, then the dealer would need to remind the scouts in seats eight and nine of the “English only at the table” rule in order to prevent them from trading secrets that fly under the radar of other players. 

There are dozens of entries in the pitching section of the scout-speak dictionary, from “command” and “control” to “arm action.” One of these buzzwords is “downhill plane,” a term that refers to pitch trajectory that has a steep slope on its approach toward the hitter. It seems to follow that pitchers who possess a high release point would induce a higher rate of ground balls. The logic behind the idea is simple enough, as anyone who has thrown a tennis ball against a wall can attest, but the statistical evidence paints a different picture.

Tall pitchers are often assumed to have exceptional downhill plane and thus produce high rates of grounders, yet the data reveals that pitcher height has no correlation with ground-ball rate among individual players. I ran the numbers for every pitcher who tossed 150 or more innings in 2011 (arbitrary cutoff alert), with a sample of 107 qualifying players and found a negligible r-value of 0.078 between pitcher height and ground-ball percentage. There are caveats with this type of analysis, given that player-height listings are about as reliable as Carlos Marmol's pitch command and that many players follow the unwritten rule of “round up and add an inch,” while the presence of just 11 data points on the Y axis creates a clustering effect, but one would expect at least a mild correlation to emerge if a strong relationship actually existed.


(Stats obtained from the
Baseball Prospectus database)

Caveats aside, there is no apparent statistical evidence to suggest that greater height is linked to more grounders, and though a pitcher's arm slot is a confounding variable in the equation for downhill plane, research conducted for “Arm Action” casts serious doubt as to the impact of release-point height on generating ground balls. *The relentless stats crew at BP advanced the research to another level, using PITCHf/x data for release-point height and running the numbers with ground-ball rates for the entire sample, finding a weak negative correlation of insignificant strength (r = -0.13, p = 0.18). The point is further solidified through anecdotal evidence, as there are countless examples of major-league pitchers whose skill sets have defied conventional wisdom, yet the adherence to old-school theories persists to this day. 

The tallest pitcher in MLB history is the Mets' Jon Rauch, whose career ground-ball rate of 35.4 percent is nearly two deviations below the mean (for the statheads, it's a z-score of -1.83). The third-tallest hurler of all time is former Padre Chris Young, whose career ground-ball rate of 29.9 percent is lower than any pitcher in the 2011 sample of 150-inning guys. Arizona's Josh Collmenter stands 6'2”, but his extreme over-the-top delivery exemplifies a common coaching instruction in the quest for downhill plane—his ground-ball rate is nearly a dead match for Rauch at 35.5 percent. Jered Weaver's 11 o'clock arm-slot was on display in Pitchology 101, and the 6'7” right-hander had just a 33.9 percent grounder rate in 2011, good for the lowest mark of any hurler with at least 150 innings on the back of his baseball card.



Such examples are offset by “diminutive” (read: my height) pitchers with ground-ball rates above 50 percent, including Tim Hudson, Ricky Romero, and Johnny Cueto. Listed at 5'10”, Cueto was the shortest player in the sample, but his 2011 ground-ball rate of 55.4 percent was good enough to rank in the top 10. Then there is the 6'6” Justin Masterson, a tall pitcher with a sidewinder arm slot and a low center of gravity that act to mute his downhill plane. Nonetheless, Masterson has generated grounders on 57 percent of balls in play during his career.

Chad Bradford was probably the most extreme counterexample to the tall pitcher paradigm. The knuckle-dragging Bradford had possibly the lowest release point in the majors, but he was able to convert frisbees into so many worm burners that his comment in BP2003 quipped, “In the future, as part of the Commissioner’s strategy to speed up games, opposing right-handed batters will be permitted to simply throw a one-hopper down to Eric Chavez rather than actually execute their plate appearance against Bradford.”

I skipped number two on the all-time height list because the man deserves his own paragraph. Randy Johnson had a much lower release point than fellow beanpoles Rauch and Young, due to excellent posture and a low slot that approached sidearm, but the Cy Young southpaw managed a 43 percent ground-ball rate that was within spitting distance of league average. Johnson stood 6'10” and exploited his long levers to gain precious inches of extension at release point, effectively shortening the distance from hand to target and allowing his already-elite radar gun velocity to play up even further. Johnson's strong momentum and long stride combined with excellent posture and a solid glove to further maximize the depth of his pitch release. The technique also allowed the Unit's slider to break closer to the plate, which acted to shrink the narrow window for hitters to recognize the pitch.

The fundamental flaw in the “get-on-top” theory is an emphasis on the initial trajectory despite an outcome that is predicated on the angle of impact. A pitcher's stuff has much greater influence on ground-ball rates than his height, ranging from the steeper trajectory of breaking pitches to the ability to upset batter timing, coaxing a hitter into topping the ball. Above-average height is a fundamental advantage, but the model breaks down on a case-by-case basis due to elements of pitcher signature.

The angle of shoulder abduction alone can account for 20 or more inches of variation in a pitcher’s release point height, and given that each player’s biological arm slot is unique, it becomes extremely difficult to alter a pitcher’s release-point height without tampering with his balance and posture. Complicating matters is the functional tradeoff that exists between release-point distance and height, as pitchers who sacrifice posture in order to “get on top of the ball” will cost themselves approximately two inches of distance for every one inch of inappropriate head movement.

Release-point distance plays a critical role in the theoretical construct of pitch velocity, with no fewer than three distinct types of velocity having been identified. The most basic form of velocity is the reading on a radar gun, which is sometimes referred to as “real velocity.” The second type of velo relates to the advantage that is earned by deep-release pitchers such as The Big Unit, with “perceived velocity” increasing along with the distance from the rubber at pitch release. A pitcher who releases the baseball closer to the plate will effectively shrink the time a batter has to react to the pitch out of hand. These pitchers are often described as “sneaky fast” due to the deceptive feeling invoked by a modest fastball that jumps on the hitter. 

The third and final category of pitch-speed is “effective velocity,” which is a term coined by Perry Husband to describe the relative amount of time that a batter has to react to a pitch due to location as well as speed. The concept is based on the fact that a batter must initiate his swing earlier in order to hit the ball squarely on a pitch that is up-and-in, and can wait a bit longer on offerings that are low-and-away. In this sense, a batter would need to invoke the timing of a 95-mph fastball in order to square up and pull a “real” 90-mph pitch that was located up-and-in. Husband found a significant effect with batter performance on pitches based on effective velocity (EV), such that hitters enjoy great success when two consecutive pitches fall within a narrow EV range, describing these situations as “at-risk pitches.” Crafty pitchers such as Greg Maddux have intuitively understood the EV phenomenon for years, utilizing the knowledge to engineer devastating pitch sequences that confound opposing batters, adding validity to the label of “smartest pitcher who ever lived.”

Many people have asked why today's pitchers appear to be so fragile when compared to the rubber arms of the past, and at least part of the answer lies in the popularization of pitching techniques that focus on batter weaknesses at the expense of pitch execution, creating a culture of pitchers who will sacrifice mechanics in the name of trajectory. I watch video of some of the old-school greats and marvel, as legends such as Bob Gibson and Walter Johnson* had more efficient deliveries than most of today's pitchers despite a century's worth of evolution. Perhaps if we changed the focus away from baserunners and batter angles in favor of pitch execution, then we would start to see the injury pendulum swing in the other direction.

*Yes, there is video of Walter Johnsonthe right-handed Randy Johnson of his day.

You need to be logged in to comment. Login or Subscribe
prs130
4/19
I realize we're talking about a lot of data that may or may not be available, but wouldn't it be interesting to hold "stuff" constant and see if the height of the release point affects the GB%? The problem is that using pitch type as a proxy for stuff might be inconsistent, since different pitchers' fastballs have different movement. Although it might be useful to observe GB% for fastballs rather than GB% generally. Another variable that might mute the advantage of downhill plane: Somebody like Roy Oswalt can throw a horizontal laser beam at knee level and get a called strike. Somebody with a much higher release point could through the ball through the exact same bat-impact point (knee height, 5-10" in front of the plate), and the ball would appear to the umpire as a ball, simply because the catcher's glove is that much lower when he receives the pitch. Ball one. The downhill pitcher then has to adjust the pitch location upward in order to get called strikes. So Oswalt can throw a ball through a knee-high impact point and get a called strike, whereas a high-release-point pitcher has to raise the pitch in order to get the same call. (Not sure if Oswalt is a good example of a low-release point pitcher, but it sure looks that way on TV.)
tombores99
4/23
I agree that it would be interesting to control for stuff, but as you mention, there are issues with data collection as well as a wide variance in fastball properties from player to player. My takeaway is that despite multiple attempts to find evidence for a connection between release-height and GB%, I have not yet found even a weak surface-level relationship. Further doubt was cast by research into the physics of downhill plane that we did for "Arm Action." As for Oswalt, I see what you mean for the moment of bat impact, which brings to question the subjective interpretation of what constitutes a "strike." Barry Zito used to have this problem with his curveball (a.k.a. the grandfather clock of 12-to-6), where it could be a strike for the front half of the zone but a ball for the back half (or vice versa), and his performance sometimes hinged on whether the umpire felt a half-zone strike was good enough. I wonder if there is evidence to support such a theory beyond the anecdotal, and would defer to PITCHf/x guru Max Marchi for such data. However, I don't think it's such a problem with "straight" fastballs, as the initial trajectory difference is very small given the overall flight distance of 54+ feet. Consider that all pitches are dropping somewhat due to gravity, and that one foot of extra release-height equates to approximately one degree of downhill trajectory (assuming the same release-distance and target location). One foot of height is essentially the difference between the shortest and tallest pitcher in the league before accounting for arm slot, and even the most extreme examples of release-height are separated by no more than 2.5 feet, or 2.5 degrees of initial trajectory.
prs130
4/19
Love the pitching analysis by the way. I bought your book and it's already changing how I teach pitching to Little Leaguers.
jhardman
4/19
Thanks for the inclusion of the Walter Johnson video links.
SaberTJ
4/19
Another awesome article Doug.
BeplerP
4/19
Doug, thank you, your work is consistently very thoughtful and clear. Question: Would your work achieve greater precision if you had ACCURATE heights for MLB pitchers? Second question: Can you associate statistically arm angle and pitching accuracy? Finally, does your work carry the possibility of any contribution to analyzing pitcher injury propensity? Thanks. Keep up the fascinating work.
tombores99
4/23
Thanks Peter. Allow me to answer your q's in order: 1) Yes, I would expect to achieve greater precision of results with greater accuracy of input variables. In fact, I think that all stats - including those in the box score - are limited by the shortcomings of their inputs (i.e. treating all singles as equal). Unfortunately, I lack access to such accurate data in this case, though the bad-ass BP stats crew busted out the PITCHf/x correlations to help back up the low-precision results. I owe them a huge debt of gratitude. 2) Not to be a broken record here, but I am afraid that the MLB data is insufficient to quantify arm angle. Setting aside the distinction between biological arm slot (abduction) and functional arm slot (face of a clock), there is no way to accurately measure a pitcher's arm slot without more sophisticated tools. We need to quantify arm slot, and finding that angle would mean tracking the pitcher's center-of-mass, establishing a perpendicular line through that COM, and then tracking the throwing hand at an angle relative to the perpendicular. Measuring pitch accuracy then requires tracking the catcher's glove at pitcher release point to see how close the pitch came to its intended target. 3) I will be covering injuries in a future edition of Raising Aces, and it is one area that is still in the naive stages of research even at the highest levels. We have learned a few things and have a ton of ideas, but the multitude of confounding variables makes injury detection a tough mystery to solve. To sum up, I wish that this data was being tracked on the field (it is being tracked in labs), and further wish that we had public access to such data. I have studied some of these issues extensively in the lab and yet there is still so much more to learn, which is what gets me amped to turn the next page in the encyclopedia of baseball research.
lipitorkid
4/19
Doug. I bought the book as well and I'm already working with my son on many of the concepts you teach in your book. I wish I had read it years ago. I loved the section on EV (effective velocity) and pitch location and the whole 16 MPH difference statistics. Please write an article about that and how it works with various MLB pitchers. I actually wrote "so cool" next to that part in the book and when my kid was reading that section he laughed at my footnote. Thanks again.
tombores99
4/23
Thanks so much for the kind words, and I am stoked that you are enjoying the book. I have to give all of the EV credit to Perry Husband, and his "downright filthy" series of books is a must-read if you are interested in diving deeper. The cool thing is that Perry studied the phenomenon from a batter's perspective before he realized the implications for pitching coaches, and we were just enamored with EV when he brought it to the NPA. I agree that would make a great article series, breaking down various pitch sequences based on Effective Velocity and evaluating the results. I did use EV to analyze a couple of at-bats for a World Series article over at BDD, in a piece that was actually inspired by a Q&A question to KG from SaberTJ. http://www.baseballdailydigest.com/2011/10/27/raising-aces-pitch-sequencing/ Small world, right?
SaberTJ
4/28
I somehow missed this article Doug. Thanks for the mention. Love your work.
SaberTJ
4/28
Should say missed this comment*
brianjamesoak
4/20
Great work, Doug. Hope to see you over at the Oakland ballpark event.