Every pitching coach has watched it happen. A pitcher works on a fix for weeks, nails it in drills, then reverts in the first inning of a real game. The cue was understood. The body simply did not reorganize around it. The constraints-led approach starts from a different question: instead of telling the athlete how to move, what can we change about the task so that a better movement becomes the natural answer?
Where the constraints-led approach comes from
In 1986, movement scientist Karl Newell described coordination as the product of constraints acting on the athlete [1]. He grouped them into three kinds, often drawn as a triangle:
- Individual constraints: what the pitcher brings, such as height, limb length, strength, mobility, fatigue, injury history and confidence.
- Environmental constraints: the world around the task, such as wind, temperature, light, the crowd and the mound surface.
- Task constraints: the rules, goals and equipment, such as the target, ball weight, distance, the count and time limits.
Change any corner and the movement reorganizes itself. Rob Gray, a perception-action researcher who has written extensively on baseball skill, explains the model in plain terms in How We Learn to Move [2], and Renshaw and colleagues set out the full coaching method in their 2019 textbook [3]. Of the three corners, coaches control the task most easily, which is why most constraints-led coaching starts there.
How a constraint changes the delivery
A constraint works by taking options away, and the athlete then finds a solution among what remains [2]. That is the key mental shift for coaches. The job becomes designing a problem where the old habit no longer works and a better pattern does.
Gray uses the image of a landscape. Habits are valleys the body rolls into easily, and years of throwing one way dig a deep one. A good constraint temporarily fills in the old valley by making the old pattern fail the task, so the body goes searching for another [2]. That is how a constraint can change a habit that years of verbal cues never touched.
Example: the connection ball
Gray describes a pitcher whose forearm flies open early [2]. The coach tucks a ball under the pitcher's throwing arm and gives one goal: when the ball falls, it should fall toward home plate. Nobody mentions the elbow. If the arm separates from the body too early, the ball drops out to the side. If the timing improves, it falls forward. The direction the ball drops tells the pitcher whether the search is heading the right way, with no coach standing there issuing verdicts.
Notice that the drill never holds the arm in a position. KNCT Athletics makes the same distinction with a balance-bike analogy: a good constraint is closer to a balance bike than to training wheels, because the athlete still has to solve the task instead of being propped into it [6]. That is practitioner framing, and it captures the idea well.
The four jobs of a good constraint
Gray gives a constraint four goals [2]. Use them as a checklist when you design a drill:
- Make the old pattern stop working. The current solution should fail the new goal.
- Let the athlete explore. No instructions on how to move, only what to achieve.
- Make useful information easier to feel. The connection ball makes arm timing obvious.
- Show whether the search is working. Gray calls this transition feedback: the ball falling more forward than sideways, a target hit, a radar reading.
Skill work on the calendar
Give constraint work a home in your week
Pitch AI builds dedicated Command days into your throwing program, so skill work gets its own slot instead of being squeezed in at the end of a session.
Pitching constraints you already know
Many familiar drills are constraints in disguise. Tread Athletics groups them into five types [5]: positional presets such as a pivot pickoff, obstacles placed where a bad pattern needs to go, visual aids such as strike-zone strings or tape lines, implements such as heavier or different balls, and timing changes such as a walking windup or a pause. Tread's coach argues that constraints can replace cues, which matters for remote coaching. A Tread coach also describes a walking windup taking one athlete from 88 to 94 mph; that is a single anecdote and not evidence of a typical effect.
The difference between a drill and a constraint is mostly in how you use it. A drill run as fixed choreography, with the coach correcting positions, is still instruction-led. The same drill with an outcome goal and room to explore becomes a constraint. For more examples you can run at home, see our pitching drills at home guide.
What does the research show?
The closest baseball test is a hitting study. Gray trained batters for six weeks to hit to the opposite field in a virtual environment, comparing a constraints-led group, a differential learning group (lots of random variation) and a group taught with prescriptive instructions, against an untrained control [4]. The instruction group improved its swing decisions but not its points. The random-variation group improved its points but not its decisions. The constraints-led group improved both and showed more functional variability in how it controlled the contact point [4].
That study was in hitters, in a virtual setting, from a single lab. No one has yet run a head-to-head trial of constraints-led training against instruction in pitchers. The case for pitching rests on motor learning theory, related sports research and coaching experience. It is a strong bet with direct pitching evidence still missing.
The approach also has real support from pressure research. In a pitching study by Gray, Orn and Woodman, experienced pitchers under pressure missed their targets more often, and what they were told to focus on changed how they moved [7]. Skills built on explicit technical rules tend to be the ones that crack under pressure, which is part of why many coaches lean on goals and constraints for in-game skills. Our post on whether coaches should let pitchers figure it out covers the autonomy side of this debate.
Do not forget the athlete's own constraints
The individual corner of the triangle changes too, just more slowly. Strength, mobility, limb length and injury history all shape which solutions are available to a given pitcher [2]. Two pitchers given the same constraint can, and often should, land on different deliveries. Fatigue is a temporary individual constraint: late in a long bullpen the body reorganizes, which is useful to know when you read a pitcher's last ten throws. Coaches can use that on purpose, for example with short command work after conditioning, as long as total workload stays sensible. Filming a few throws before and after a constraint block, as our post on video review for pitching mechanics describes, shows whether the delivery actually changed.
This is also why constraints pair well with physical training. A stronger, more mobile athlete has more movement options to find, and the constraint helps him find a good one.
Common misconceptions
- "The constraints-led approach means no coaching." Gray's examples all have a clear goal and feedback. Constraints replace detailed body instructions, and the coach's job moves to designing the task [2].
- "There is one perfect delivery to constrain toward." Constraints help each athlete find a solution that fits their body. That is a different aim from copying a template, and our pitching mechanics guide explains why individual solutions vary.
- "A constraint is a brace." If the tool holds the athlete in position, it is closer to training wheels and less likely to transfer [6].
- "It replaces strength and velocity work." Constraints shape skill. They sit alongside the physical work in a weekly training schedule.
Try it this week
Pick one problem you or your pitcher has been cueing for a long time. Ask which corner of Newell's triangle you can change so that the old pattern fails. Give one outcome goal, let the athlete explore for 10 to 20 throws, and watch whether the feedback (ball flight or the connection ball) is heading the right way. Part 2 covers the specific dials you can turn: how to use task constraints to improve command. Part 3 shows how to make bullpens more like games, and Part 4 covers getting a new pitch from the bullpen into games.
Frequently asked questions
What is the constraints-led approach in baseball?
It is a coaching method based on Karl Newell's 1986 model, where movement emerges from individual, environmental and task constraints. Coaches change the task, such as the target, ball or rules, so a better movement becomes the natural solution instead of relying on detailed verbal instructions.
What is an example of a pitching constraint?
The connection ball: a ball tucked under the throwing arm with the goal that it falls toward home plate. If the arm flies open early, the ball drops sideways, so the pitcher gets feedback on timing without any elbow cue.
Is the constraints-led approach proven for pitchers?
Not directly. The strongest baseball evidence comes from a virtual-reality hitting study where the constraints-led group improved more than instruction or random-variation groups. No head-to-head pitching trial has been published yet.
Does the constraints-led approach mean coaches stop coaching?
No. Coaches still set clear goals and give feedback. The difference is that they design the task to steer movement instead of describing the correct body position.
Constraints that stick
Build Command days into your program
Pitch AI builds dedicated Command days into your throwing program and tracks the intent of every set. AI Video Review gives you cues and top fixes from your own throwing video, so you can see whether a constraint actually changed your delivery.
References
- Newell KM. Constraints on the development of coordination. In: Wade MG, Whiting HTA, eds. Motor Development in Children: Aspects of Coordination and Control. Martinus Nijhoff; 1986:341-360.
- Gray R. How We Learn to Move: A Revolution in the Way We Coach & Practice Sports Skills. 2021. Chapters 4 and 7.
- Renshaw I, Davids K, Newcombe D, Roberts W. The Constraints-Led Approach: Principles for Sports Coaching and Practice Design. Routledge; 2019.
- Gray R. Comparing the constraints led approach, differential learning and prescriptive instruction for training opposite-field hitting in baseball. Psychology of Sport and Exercise. 2020;51:101797. doi:10.1016/j.psychsport.2020.101797
- Tread Athletics. "5 Types Of Constraints To Clean Up Your Mechanics." YouTube, 2026-10-11. youtu.be/cujsMwd8NcA
- KNCT Athletics. "Fixing A Pushy Arm Action With Constraints." YouTube, 2026-09-22. youtu.be/J00_uPqEXis
- Gray R, Orn A, Woodman T. Ironic and reinvestment effects in baseball pitching: how information about an opponent can influence performance under pressure. Journal of Sport and Exercise Psychology. 2017;39(1):3-12. doi:10.1123/jsep.2016-0035