Conducted by Morrie Silver and Ben Martin
Abstract
Background:
Pitch Tunneling is usually analyzed from the viewpoint behind the pitcher, with the intent of seeing how long multiple pitch types can share a similar tunnel on their path to home plate. The purpose of this study was to determine whether looking at pitch tunneling from behind the mound provides an accurate idea of how these pitches tunnel to hitters.
Hypothesis:
The batter’s viewpoint will provide a different perspective of tunneling when compared to the ‘basic’ view behind the mound.
Study Design:
Controlled laboratory study.
Methods:
With the use of three different high-speed camera angles, video was captured as pitches were thrown by a pitching machine. Four different pitch types were recorded by all cameras and pitch metric data was recorded by a Rapsodo pitch tracking unit as well as a Pocket Radar (radar gun). In this experiment, fastballs were utilized as the control variable to which other pitches were compared. The footage of the three other pitches were then individually overlaid onto the fastball footage and measured using photoshop and the pythagorean theorem. The measurements were recorded.
Results:
Pitches tunneled differently when viewed from the batters’ perspectives when compared to the pitcher’s ‘basic’ tunneling view. Furthermore, a difference was seen in how well pitches tunneled depending upon which side of the plate they were viewed from. There were significant differences in the tunneling effects based upon perspectives. It was found that if a pitch appears to tunnel from behind the mound, it will not necessarily tunnel from the batter’s perspective. Alternatively, pitches that do not tunnel from behind the mound, may in fact, tunnel from the batter’s perspective.
Conclusions:
The difference in the tunneling effect based on perspective provides support for the hypothesis. While prone to the subjectivity of analysis, there is still validity to the facts found.
Background:
Pitch tunneling has become all the rage in pitching development within the last few years. With the new accessibility to cameras and video footage of pitchers and their pitches, there has been an overhaul in how pitchers have started to develop. The concept of designing pitches and building velocity has come to the forefront of the common baseball fan (and players) attention. The principles of pitch design play part to a bigger training idea; building an arsenal. A properly built-out arsenal is a group of pitches which work well together. But what does working well together mean? Pitches that work well together usually mean that they’re different pitch types that can mirror each other for as long as possible and then disperse into multiple, different locations. This allows for more information denial to the hitter as now multiple pitches look similar as the pitch approaches the hitter’s decision point on its path to home plate. Perry Husband, famed baseball scientist, coined the idea of pitch tunneling and informed pitchers that if they could keep their various pitch types on the same path (“or tunnel”) for at least the first 20 feet of flight before separating, the hitter's best hopes are “guesswork and luck”.
Pitch tunneling is usually analyzed by a camera placed behind the pitcher. The purpose of this study was to determine whether looking at pitch tunneling from behind the mound provided an accurate idea of how pitches tunnel to hitters.
As seen through MLB.com’s Baseball Savant page, the user can view the pitcher’s arsenal through the use of 3D Pitch Tracks flight data from different perspectives; that of the right-handed hitter, the left-handed hitter, and behind the pitcher. By switching to different viewpoints, the difference in how pitches tunnel becomes apparent. It is this concept that inspired this study.
Below are four illustrations of two different pitchers.
The arsenals for both pitchers are analyzed from the three different angles as discussed in this study. It is the difference in these three angles that illustrate the concept in this study. If there is a clear difference in how the pitches tunnel when utilizing this objective data, there is reason to believe that the ‘basic’ behind-the-mound view is insufficient when trying to understand if pitches truly tunnel.
A Word about Trevor Bauer
As one of the most innovative minds in baseball, right-handed big league All-Star pitcher (Alexa, play “Taste” by Tyga) Trevor Bauer has long been aware of the concept of tunneling. In fact, he and his father, Warren, engineered a training constraint that would help him achieve pitch tunneling at a high level. The two built a metal frame out of rebar with a 13 by 10 inch opening. If Bauer's practice pitches passed through the hole, he knew they were in the tunnel. More can be found about Bauer’s contraption and training method in Jason Turbow’s 2014 piece on SBNation.com where he interviewed Perry Husband.
A Word about The Parallax Error
When trying to analyze anything in two-dimensions that is actually three-dimensional, the parallax error cannot be ignored. Perhaps one of the biggest factors at play here is the difference in perception based on different perspectives. Below is a tweet, explaining the issues that can arise if not acknowledged. Seattle-based player development company, Driveline, explains it well, “parallax error means that if you are trying to measure the same thing from different angles, there will be different measurements simply because of the different camera angles.”
In the accompanying tweet, there are two pitches being thrown. One with more movement than the other, though you cannot tell the difference due to the parallax error.
The parallax error is the driving force behind this study. While so many analyses of the game of baseball come by way of video, viewers are left having compared events from different camera angles. This is acknowledged when it comes to analyzing the biomechanics of the swing or pitching, but ignored when discussing the effects of pitch tunneling.
A Word about Gaze Tracking
Beginning with Hubbard and Seng in 1954 and now continuing most notably with Driveline, the concept of gaze tracking and trying to understand how hitters see the ball is being studied. It was important to understand the work being done in this field, as part of this study reflected how a hitter sees pitches. In Rob’s (no last name) Pitch Tunneling and Perceptually Equivalent Pitches blog post, he writes, “When we are talking about things related to a batter’s vision like tunneling we need to analyze things in terms of perceptual information not physical variables like feet and mph.”
As reiterated throughout Driveline’s Gaze Tracking article, the concept that hitter’s cannot see the ball in the last 150 milliseconds shows that the closer the two pitches can be in relation to each other, the less perceptual information pitchers are giving the hitter. The less information the hitter can gain from looking at pitches, the less accurately they can predict where the ball will go.
Further progress in this field of study will continue to provide information to pitching development as it will uncover the effects of pitch tunneling.
A Word about Subjective Analysis
While this study includes objective measurements, the majority of this is based on the premise of subjective analysis. Although subjective analysis is often deemed biased or less valuable than objective, this study focuses on the perceptual effects of perspective, thus subjective analysis provides value.
A Word about Sequencing
After reading this study and understanding the emphasis placed on tunneling, the question arises as to where it falls into pitching strategy. Pitch tunneling and pitch sequencing are often viewed as two separate entities on their own, but each is rendered significantly less effective without the other. In Josh Kalk’s 2009 Pitch sequence; High Fastball then Curveball, and subsequently Pitch Sequencing, the importance that the previous pitch has on the next pitch can be seen. Kalk goes through an analysis of pitch types and different pitches that followed and saw which performed well and which didn't (using a slightly modified version of Runs100). Kalk doesn't mention the word ‘tunneling’ once in his work (probably because it wasn't coined until Perry Husband discussed this novel idea in a SB Nation article in 2014), but rather talked about the pitch coming out of a different arm slot or having a “hump” to it. Kalk explains how having less of a hump to your curveball could help deceive the hitter. So while he didn't mention the term ‘tunneling’, Kalk is on record as understanding the unbirthed concept.
Dan Blewett’s outstanding 2017 work in The Hardball Time, entitled, Tunneling: Is it Real? And how do Pitchers actually Pitch? should also be noted. In it, he says “pitch tunneling is most relevant as a concept of pitch sequencing”. He continues, “Though pitch tunnels are intriguing and can certainly aid pitchers in making good pitch-calling decisions, expected outcomes will likely remain the gold standard.”
This analysis can be put into simple old-school baseball terminology. Blewett is basically saying that a pitcher is less likely to get beat by the wrong pitch in the right location, than the right pitch in the wrong location. Conversely, the margin of error could be larger on the right pitch (tunnels well with the previous pitch) in a bad location than on the wrong pitch (doesn’t tunnel well with the previous pitch) in a good location.
Research Question:
1. Do pitches traditionally thought to tunnel from behind the mound, tunnel to the hitter as well?
a. Is there a difference between how certain pitches tunnel from each perspective (right-handed hitters vs. left-handed hitters)?
Methods:
The study was conducted in a controlled environment, using a batting cage with a portable turf pitching mound and an ATEC Casey Pro Pitching Machine. The machine was placed on the mound and measured to match the Major League Baseball average release point for right-handed and left-handed pitchers. A three-camera setup was utilized. Three cameras were able to provide the perspectives of right-handed batters, left-handed batters, as well as the ‘basic’ pitch tunneling view from behind the mound. Two Sony RX 100 (mark VI) cameras were used for the two vantage points in the batter’s boxes, as well as one Edgertronic camera behind the mound for the ‘basic’ tunneling view. All three cameras shot at 960 frames per second. Both batter’s box cameras were slightly turned to mimic the hitter’s gaze while standing in the box (avoiding a straight on view of the pitch) and were placed upon custom built tripods. These tripods were constructed to match a hitter’s eye level in the batter’s box. All three cameras had focal points in the frame to ensure the same frame was captured for all pitches to provide uniformity. The batter’s box cameras were controlled remotely as a way to avoid nudging the camera even the slightest which would have produced a less consistent frame to work with and ultimately, a poor overlay. In addition to the three camera setup, a Rapsodo Pitch Tracking 2.0 unit was also used. This was to provide further verification of speed gauging as well as to capture other pitch metrics such as spin rate, vertical and horizontal break, tilt, etc., to ensure pitches were as game-like as possible. For example, the machine was adjusted to mimic the true average tilt for those pitches.
Experiment Setup
A Word about our Failures
It should be noted that, in the first attempt, two different camera tripods were used. After capturing one pitch, the cameras were removed and markers were placed on the ground in an attempt to put the tripods back in the right place to capture the same frame. Not only was this a total failure as far as replacing them in the identical spot, but it also made it impossible to ensure the tripods were consistent and level with each other due to the slight variance in the two different brands of tripods.
Consistency Noted:
Another note that should be considered is the fact that for all three camera angles, one pitch was captured and recorded for each event. Simply put, the right-handed hitter’s camera recorded the same pitch as the left-handed hitter’s camera, as did the ‘basic’ behind the mound camera. It was vital to ensure that all three cameras were viewing the exact same pitch and not a similar one or something with similar metrics. The exact pitch needed to be captured by three different cameras. This is where a lot of time was consumed in testing, being that multiple times, two out of three cameras would capture the pitch while the third one erred, only to have the same thing happen on the next pitch with a different two out of three capturing the pitch. Three different pitches could have been pieced together, but this would have provided inconsistent data.
An additional area worth noting under this section is the pitching machine that was used. While the pitching machine used proved to be effective, at times it took multiple takes to get a suitable pitch as the variability in the machine would result in pitches being thrown outside of the strike zone.
Procedure:
After setting up all of the equipment, the first step in the study was setting up the pitching machine for a right-handed fastball.
Both the right-handed and left-handed fastballs were set to the same angle throughout the study, but it should be noted that due to the requirement of retesting certain pitches, there was slight velo fluctuation. This does not impact the findings, as all overlays were done within the same velocity and the same clip. Simply, a 97 mph fastball is never compared to a 90 mph fastball. Throughout the study, the same pitch event is analyzed throughout all three camera angles. Refer to the paragraph entitled Consistency Noted for further clarification and explanation.
Note on Fastballs
Capturing a suitable fastball was a high priority, since the fastball was used as the control variable for all pitches from their respective handedness. In many popular tunneling videos, the pitches paired are typically a fastball and an off-speed pitch. While fastballs were used to see how well off-speed pitches tunneled, it was also a frame of reference with respect to common tunneling videos.
The procedure for collecting video of each pitch was as follows:
Set up protective netting in front of cameras (see Setup Note)
Adjust pitching machine to ensure desired velocity, tilt, location.
Test pitches until desired results occur
Remove protective netting
Ensure cameras on standby mode
Communicated verbally and count down from 3 to drop of pitch into machine, to which all three cameras click record on count of 1
Confirm complete recording of video and confirm the desired pitch was achieved
If no confirmation, redo steps 5-9.
If confirmation is upheld, move on to the next pitch starting with step one and repeat.
Setup Note
In order to ensure the safety of the cameras, and to make sure they were not altered while adjusting the pitching machine, protective netting was used. Two nets were kept in front of the cameras until the pitching machine had the next pitch type dialed in for a strike, and the cameras were no longer in any danger of being hit.
Measurement Note
Distance measurements are included in this study, for the purposes of comparative analysis. These measurements are not intended to be exact in the 3-D realm.
Results:
FB/CB - RHP
FB/CB - LHP
FB/CB SUMMARY:
Comparing the two fastball/curveball overlays, the left-handed FB/CB appears to tunnel tighter than the right-handed FB/CB overlay, based on the 'basic' behind the mound view. When looking at both side-by-side views of the overlays, it would be tough to distinguish the difference between the two pitches. One distinction that can be made of the left-handed pairing, viewed from the left-handed hitters’ perspective, is the noticeable ‘hump’ on the curveball. While not seen from the ‘basic’ behind the mound view, it can be seen from the LHH perspective.
Concisely stated, our hypothesis is confirmed as the three different viewpoints prove to provide different visual information.
FB/SL - RHP
FB/SL - LHP
FB/SL SUMMARY:
Comparing the two fastball/slider overlays, it is evident that the pitches both seem to ‘tunnel’ from the ‘basic’ behind the mound view, but effectively tunnel differently based on the batter’s perspective. From the same handedness, the pitches tunnel better. For example, the left-handed FB/SL tunnel is tighter when viewed from the left-handed hitters perspective, than when compared to the right-hand hitters perspective, and vice versa.
Concisely stated, our hypothesis is confirmed as the three different viewpoints prove to provide different visual information.
FB/CH - RHP
FB/CH - LHP
FB/CH SUMMARY:
Comparing the two fastball/changeup overlays, it is evident that the handedness of the pitcher plays a factor with this pairing. The same handedness provided the appearance of a tighter tunnel.
It should be noted that changeups rely more on a speed differential, and aren't always evaluated on how far they deviate from the fastball’s path.
Concisely stated, our hypothesis is confirmed as the three different viewpoints prove to provide different visual information.
Conclusion:
This study provided support for the hypothesis that pitches appear to tunnel differently from the hitter’s perspective when compared to the ‘basic’ tunneling view. Additionally, it provided support for the hypothesis that pitches tunnel differently based on the handedness of the batter. The study showed some pairings tunnel more tightly while others tunnel more loosely. If tunneling is something that is being sought after as a training and development goal for a pitcher, it is less important for that pitcher to evaluate him/herself from the ‘basic’ tunneling view and more important to evaluate himself from the batter’s box views.
Limitations
The study was confined to the capabilities of a two-wheeled pitching machine which, in itself, is limited. Collecting a large amount of footage and data from all three angles with a live human pitcher would provide more robust results, though a pitcher with the ability to throw all four pitch types, for the sake of research, would be a challenge.
Further Research:
One area where further research could be explored is how different pitches tunnel in different locations. While this study compared three types of pitches with a fastball with high vertical break near the top of the strike zone, it is worth researching how other pitch types tunnel in different locations. For example, how a fastball with low vertical break, thrown low in the strike zone, tunnels with a slider, or how a curveball and a changeup tunnel with each other. Additionally, replicating this study using a live pitching could provide valuable information, though finding a pitcher with the ability to throw all four pitch types, for the sake of the research, would be a challenge.
Special Thanks:
We would like to extend a thank you to Matthew Toohey for his assistance with the data collection for this study.
Works Cited
Aucoin, D. (2020, May 21). How do Batters See the Ball? A Review of Gaze Research in Batting. Retrieved July 3, 2020, from https://www.drivelinebaseball.com/2019/02/batters-see-ball-review-gaze-research-batting/
Aucoin, D. (Director). (2019, October 10). Pitch Tunneling Part 1: Establishing a Definition [Video file]. Retrieved July 1, 2020, from https://plus.drivelinebaseball.com/pitch-tunneling-part-1-establishing-a-definition-0840/
Aucoin, D. (Director). (2019, October 10). Pitch Tunneling Part 2: A Dive Into Current Research [Video file]. Retrieved July 1, 2020, from https://plus.drivelinebaseball.com/pitch-tunneling-part-2-a-dive-into-current-research-1250/
Aucoin, D. (Director). (2019, October 10). Pitch Tunneling Part 3: Limitations of What We Know [Video file]. Retrieved July 1, 2020, from https://plus.drivelinebaseball.com/pitch-tunneling-part-3-limitations-of-what-we-know-1626/
Blewett, D. (2017, June 16). Pitch Tunneling: Is It Real? And How Do Pitchers Actually Pitch? Retrieved July 7, 2020, from https://tht.fangraphs.com/pitch-tunneling-is-it-real-and-how-do-pitchers-actually-pitch/
Kalk, J. (2009, February 10). Pitch sequencing. Retrieved July 4, 2020, from https://tht.fangraphs.com/pitch-sequencing/
Kalk, J. (2009, February 3). Pitch sequence: High fastball then curveball. Retrieved July 5, 2020, from https://tht.fangraphs.com/pitch-sequence-high-fastball-then-curveball/
No-Last-Name, R. (2017, June 24). Pitch Tunneling & Perceptually Equivalent Pitches. Retrieved July 3, 2020, from https://perceptionaction.com/pitchtunnels/
Pavlidis, H., Judge, J., & Long, J. (2017, October 25). Prospectus Feature: Introducing Pitch Tunnels. Retrieved July 4, 2020, from https://www.baseballprospectus.com/news/article/31030/prospectus-feature-introducing-pitch-tunnels/
The Pitfalls of 2D "Biomechanical" Analysis. (2018, February 16). Retrieved July 3, 2020, from https://www.drivelinebaseball.com/2018/02/pitfalls-2d-analysis/
Spörri, J. (2018, August 01). Figure 14: Direct linear transformation (DLT) method (own illustration... Retrieved July 2, 2020, from https://www.researchgate.net/figure/Direct-linear-transformation-DLT-method-own-illustration-based-on-Abdel-Aziz-and_fig14_270049817
Turbow, J. (2014, June 18). The Essence of Velocity. Retrieved July 4, 2020, from https://www.sbnation.com/longform/2014/6/18/5818380/effective-velocity-pitching-theory-profile-perry-husband
Comentários