Every pitching coach has said it, and every pitcher has heard it after a walk: repeat your delivery. The intent behind the cue is good. Command depends on getting the ball out of the hand the same way, pitch after pitch. The problem is the assumption hiding underneath it, that the body should copy one ideal motion like a machine stamping parts.
Biomechanics research on skilled pitchers points somewhere else. Their joints vary from throw to throw, and those variations are arranged so they cancel each other out at release. What repeats is the outcome. Understanding that difference changes how you should practice command.
Can a pitcher repeat the exact same delivery?
No, and the numbers show why a perfect copy would not be enough even if it were possible. Kusafuka and colleagues tracked 187 pitches from seven skilled pitchers and fed the release data into a ball flight model [1]. A change of one degree in vertical release angle moved the pitch about 30 cm (roughly a foot) at the plate. A 1 m/s change in ball speed moved it about 20 cm vertically. Shifting the release point itself by 1 cm moved the pitch only about 1 cm.
Think about what that means. Location is far more sensitive to the angle the ball leaves the hand than to where the hand is in space. A one degree error is invisible on video. No human nervous system can reproduce dozens of joint rotations closely enough to hold release angle inside that window by copying alone. Fatigue, mound slope, a slick ball and adrenaline all change the starting conditions of every throw. Something else has to be keeping the ball on target.
How skilled pitchers stay accurate without a perfect copy
That something is compensation. Matsuo and colleagues measured twelve semi-professional pitchers throwing fastballs at about 37 m/s [2]. Each pitcher's joint angles varied from throw to throw. When the researchers mathematically shuffled those joint values to remove the way they moved together, the hand at release became less consistent. In the real throws, the variations were coordinated so they offset each other and kept the hand steady. Interestingly, the stride leg contributed more to that stabilizing than the throwing arm did.
The same thing shows up at the level of the ball. Kudo and colleagues had participants practice an accuracy throw for 150 trials [3]. Accuracy improved mainly because the release parameters (speed, angle and height) learned to compensate for each other, even though no single parameter ever became perfectly consistent on its own.
Command comes from parts that correct each other in real time. A delivery that looks slightly different on video can still produce the same release angle, and that is what the hitter sees.
What "repetition without repetition" means for pitchers
The Russian scientist Nikolai Bernstein noticed this pattern decades ago while studying expert blacksmiths [4]. The hammer landed on the same spot of the chisel every time, yet the path of the arm varied more than the path of the hammer. The joints were correcting each other on the fly. Bernstein described practice as "repetition without repetition": you repeat the solving of the problem, and each solution is a little different.
Rob Gray, a perception and motor learning researcher, uses this as the starting point of his book on coaching [5]. His summary is useful for pitchers: what happens after the movement starts matters as much as the plan. Skill is the ability to make the right adjustment in the moment without having to think it through.
Good variability vs bad variability
Sloppy mechanics still lose, though. Fleisig and colleagues compared 93 pitchers from youth to the major leagues [6]. Within each pitcher, variability in six of eleven kinematic measures (including foot placement, knee flexion, elbow flexion and trunk forward tilt) was greatest in youth and lower at higher levels. Release consistency also matters in the big leagues. In 190 MLB starters and about 76,000 pitches, variation in release location was a significant predictor of fielding independent pitching (FIP) [7], and MLB pitchers show less release point variability than minor leaguers [8].
The way to reconcile these findings comes from motor control research on the uncontrolled manifold [9]. Some variation pushes the outcome off target. Call it bad variability. Other variation is arranged so that it cancels out, leaving the outcome where you wanted it. Call it good variability. Elite pitchers strip out much of the first kind and keep, or even rely on, the second.
Driveline reports a version of this pattern in its own lab data from 27 college to MLB athletes [10]. Better command went with more consistent positions at foot plant (glove side shoulder and torso tilt) and with more variability later in the delivery, near layback and in pronation at release. This is company data and has not been peer reviewed, so treat it as a practitioner observation. It fits the research well: stable early, adaptable late.
Practice the adjustment
Put command work on the calendar
Pitch AI builds dedicated Command days into your throwing program, so moving targets, windup and stretch work, and charted misses have a regular place in your week.
How to train command that adapts
If command is the ability to find the release again under changing conditions, practice should give the body that job regularly. Here are practical ways to build it into bullpens and flat ground work, drawn from Gray's recommendations on variable practice [5]:
- Move the glove every pitch. The catcher sets up in a new spot for each throw so you never get two identical aiming problems in a row.
- Mix windup and stretch. Alternate, or let the catcher call it. Each start position changes timing slightly, and your release has to adjust.
- Change the mound. Throw a pen on an unfamiliar mound or at a different field when you can. Spring training stories about pitchers sailing their first throws often trace back to a winter of identical indoor conditions.
- Chart how far you miss. A ball or strike call tells your body very little. A note like "six inches arm side, slightly up" gives it something to correct.
The last point has support in pitching research. Kusafuka and colleagues followed 14 skilled pitchers across 30 pitches each and found they corrected release angle from one pitch to the next, with corrections that depended on the direction of the previous miss [11]. Pitchers with very large release angle variability made fewer horizontal corrections. Error correction is a real part of command, and it runs on information about the miss.
A sample command pen for this week
For a pitcher who can already throw strikes with two pitches: 30 pitches total. The catcher calls the pitch and moves the glove after every throw. Alternate five from the windup and five from the stretch. Someone charts miss distance in inches and direction. Note the average miss for the first ten and the last ten. Next time out, compare your first ten pitches to today's. That comparison tells you what stuck, which matters more than how the pen felt.
Learn your own miss pattern
Misses have a shape. Shinya and colleagues found that each pitcher's location errors form an ellipse whose long axis follows the arm path [12]. For a right handed overhand pitcher, that usually means misses spread along a line from up and to the arm side toward down and to the glove side. Once you know the shape of your pattern, you can choose targets that keep the long axis of the ellipse inside the zone, and you can judge progress by whether the ellipse shrinks.
Grade yourself in the conditions where command usually breaks: late in a pen, from the stretch, on a new mound. Consistency in those moments is the real test.
When to keep things simpler
Variable practice has a dose. Gray's rule of thumb is to add variability in inverse proportion to how variable the athlete already is [5]. A young pitcher who sprays the ball everywhere already has plenty of variation; adding chaos on top of that rarely helps. Start that pitcher with a big target and a stable setup, then add changes as strike rates climb. For an experienced pitcher who looks perfect in clean bullpens and struggles in games, the opposite is true. That pitcher needs more variety, sooner. Our posts on youth pitcher development and pitching mechanics that actually matter cover the physical side of this.
The cue "repeat your delivery" still has a place, as shorthand for a stable setup and foot plant. Just aim it at the result. You want to repeat the release and the ball flight, and the body will find a slightly different route to get there each time. Part 2 of this series explains what dexterity is and how pitchers build it, and you can see the practice schedule side in random vs blocked practice for pitchers.
Frequently asked questions
Should a pitcher try to repeat the exact same delivery?
A pitcher should aim to repeat the release and the result. Research on skilled pitchers shows their joints vary from throw to throw, and those variations cancel each other out so the hand and release angle stay consistent. A stable setup and foot plant help, but chasing an identical motion is impossible and misses the point.
How much does release angle affect pitch location?
In a study of skilled pitchers, a one degree change in vertical release angle moved the pitch about 30 cm, roughly a foot, at the plate. A one centimeter change in release point moved it only about one centimeter. That is why command depends on coordinating many body parts to protect release angle.
Is variability in pitching mechanics bad?
Some is and some is not. Variation that pushes the ball off target hurts command, and higher level pitchers show less of it. Variation that offsets other variation and keeps the release steady is part of how skilled pitchers stay accurate.
What drills improve pitching command?
Variable command work: the catcher moves the glove every pitch, you alternate windup and stretch, you throw on different mounds, and someone charts how far each pitch misses. Keep targets larger for young or very wild pitchers and add variation as strike rates rise.
Command that travels
Train command that holds up anywhere
Pitch AI builds dedicated Command days into your throwing program and logs every throwing session, so the variable work in this article becomes a weekly habit you can see adding up.
Sources and further reading
- [1] Kusafuka A, Kobayashi H, Miki T, et al. (2020). Influence of release parameters on pitch location in skilled baseball pitching. Frontiers in Sports and Active Living, 2, 36. https://doi.org/10.3389/fspor.2020.00036
- [2] Matsuo T, Jinji T, Hirayama D, et al. (2020). Consistent hand dynamics are achieved by controlling variabilities among joint movements during fastball pitching. Frontiers in Sports and Active Living, 2, 579377. https://doi.org/10.3389/fspor.2020.579377
- [3] Kudo K, Tsutsui S, Ishikura T, et al. (2000). Compensatory coordination of release parameters in a throwing task. Journal of Motor Behavior, 32(4), 337-345. https://doi.org/10.1080/00222890009601384
- [4] Bernstein NA (1967). The Co-ordination and Regulation of Movements. Pergamon Press.
- [5] Gray R (2021). How We Learn to Move: A Revolution in the Way We Coach and Practice Sports Skills. Perception Action Consulting and Education (chapters 1, 8 and 9).
- [6] Fleisig G, Chu Y, Weber A, Andrews J (2009). Variability in baseball pitching biomechanics among various levels of competition. Sports Biomechanics, 8(1), 10-21. https://doi.org/10.1080/14763140802629958
- [7] Whiteside D, Martini DN, Zernicke RF, Goulet GC (2016). Ball speed and release consistency predict pitching success in Major League Baseball. Journal of Strength and Conditioning Research, 30(7), 1787-1795. https://doi.org/10.1519/JSC.0000000000001296
- [8] Wakamiya K, Nagamoto H, Yamaguchi R, et al. (2024). Relationship between ball release point variability and pitching performance in Major League Baseball. Frontiers in Sports and Active Living, 6, 1447665. https://doi.org/10.3389/fspor.2024.1447665
- [9] Scholz JP, Schöner G (1999). The uncontrolled manifold concept: identifying control variables for a functional task. Experimental Brain Research, 126(3), 289-306. https://doi.org/10.1007/s002210050738
- [10] Pelletier C (2026). The interaction of biomechanics and command. Driveline Baseball (company blog, not peer reviewed). https://www.drivelinebaseball.com/2026/02/the-interaction-of-biomechanics-and-command/
- [11] Kusafuka A, Okegawa T, Yamamoto R, et al. (2025). Two-dimensional trial-by-trial error correction for accurate baseball pitching. Scientific Reports, 15, 12300. https://doi.org/10.1038/s41598-025-97146-5
- [12] Shinya M, Tsuchiya S, Yamada Y, et al. (2017). Pitching form determines probabilistic structure of errors in pitch location. Journal of Sports Sciences, 35(21), 2142-2147. https://doi.org/10.1080/02640414.2016.1258484