The standard bullpen looks the same in most programs: catcher sets up in the middle or on a corner, the pitcher throws a block of fastballs, then a block of breaking balls, at an easy count with nobody in the box and no clock. It is comfortable and repeatable, and it trains a version of command that rarely exists in games. In Part 1 we covered why changing the task changes the movement. This post is the practical toolkit.
What is a task constraint?
A task constraint is any rule or piece of equipment that shapes how a throw has to be made [2]. Coaches can change task constraints in seconds, which makes them the easiest lever in Newell's model [8]. Each one has an easier end and a harder end, so you can scale the difficulty to the pitcher in front of you.
The dials on your bullpen
| Dial | Easier | Harder |
|---|---|---|
| Target size | Whole strike zone | One quadrant, edges only, a small cone |
| Location call | Same spot for a block | Called late, just before the set |
| Pitch order | Blocked (10 fastballs, then 10 sliders) | Random pitch calls |
| Count and situation | 0-0, no consequence | 3-1 with a runner on, points at stake |
| Clock | None | Slightly shorter than the game clock |
| Distance | One fixed distance | Varied distance between pitches |
These dials come from Gray's constraint examples [2] and Driveline's command programming [1]. Turning a dial toward the hard end usually makes practice look worse in the moment. That is expected. In a classic baseball study, college hitters who practiced against a random mix of fastballs, curveballs and changeups improved 56.7% from pretest to a random transfer test, compared with 24.8% for a blocked-practice group and 6.2% for controls [6]. That was hitting, and it remains the closest baseball evidence for randomizing pitch types and locations. We cover that research in more depth in random vs blocked practice for pitchers.
How hard should it be?
Guadagnoli and Lee's challenge point framework says learning depends on matching task difficulty to the performer's skill: too easy gives little to learn, and too hard overwhelms [5]. A practical reading is to keep success rates clearly below 100% but well above zero, and to shrink the target or add a count once a pitcher starts hitting it most of the time. The exact cut-offs are coaching judgment.
Change the ball
The ball itself can be a constraint. Driveline's Differential Command Program uses 4, 5 and 6 ounce balls plus balls about 5% larger and smaller than regulation, thrown first in blocks and later in random order, and moves a pitcher up once he throws at least half strikes [3]. Their command programming in Hacking the Kinetic Chain describes similar blocked and random weighted-ball command days [1]. In a recent video, Driveline adds softballs, handballs and moving targets to the same idea and describes its best-command pitchers as consistent into foot plant but more adaptable at release [7]. That last point comes from Driveline's internal lab data and has not been published.
Treat ball variation as a smart tool that still lacks published outcome data for pitching command. One practitioner nuance is worth knowing. KNCT Athletics suggests matching the tool to the task: heavier balls tend to drive an earlier arm flip-up, so if the goal is cleaning up timing, a lighter ball may fight the athlete less [9]. If you use weighted balls, the stress side matters too, and our weighted ball research guide covers dose and arm load.
Track what changed
Log every grip change in Pitch Lab
Pitch AI's Pitch Lab logs every grip change, so you can see what it did to the shape. It also records velo, IVB, spin, axis and release for each pitch and plots them on a movement chart.
Add a movement rule, then aim
A second family of constraints puts a rule on the movement itself:
- Throw from the stretch only.
- Land on a soft or sandy spot.
- Keep a connection ball under the throwing arm.
- Start every delivery from a set position or a pause.
Gray describes a sandy landing area used with a pitcher who rolled the front ankle at landing: the soft surface made a small flaw impossible to ignore, and the pitcher had to search for a new solution [2]. The rule reshapes the movement. Then add a target. The target keeps the throw honest, because a delivery change that ruins location is not a usable change. Start with the rule alone. Once it feels natural, add a big target and shrink it over time.
Give the ball a job: shape goals
With a ball-tracking device, ball flight itself becomes the constraint. Ask for a specific shape, say 15 inches of horizontal break on a sweeper, and let the pitcher search for the grip and release that create it. Brian Bannister, speaking on a Tread Athletics podcast, describes his pitch design sessions as a lab where pitchers are free to experiment, and stresses that sessions meant to match game shapes need near-game intent [4]. Without tracking, you can still give a shape goal with a camera and a string or target to break across.
Shape work pairs naturally with command work. Once the shape shows up in most throws, add a location goal and the pitch moves into the next stage. Our pitch design workflow covers the measurement side, and the sweeper guide has grips and cues for that specific shape.
Constraints in catch play
You do not need a mound to start. Catch play is the highest-volume throwing most pitchers do, and it is usually the least designed. On Rubicon's "Casting the Die" podcast, the host argues that many pitchers throw every catch-play ball as a low-intent fastball, and suggests manipulating ball flight at distance, including long tossing a slider [10]. That is practitioner opinion, and it turns idle volume into skill work.
Coaches disagree on the details, which is worth knowing. Driveline suggests trying to hit your partner in the chest on every throw [7], while KNCT Athletics argues that a chest-only target encourages steering the ball [9]. Both camps agree catch play should have a goal. Pick one per session and change it from week to week.
The recipe for any constraint session
- Pick one problem. For example, misses arm side when the front side opens early.
- Choose a constraint that makes it obvious. Gray calls this making the problem louder [2]. A glove-side target that the open-front-side miss cannot reach does the job.
- Give an outcome goal instead of a body instruction. "Hit the glove-side half" beats "stay closed."
- Give feedback on the search, and adjust difficulty. Ball flight and target hits tell the pitcher whether he is heading the right way. Make the task harder as success rises and easier if it collapses [5].
A sample 30-pitch command bullpen
Here is one way to turn the dials in a single session for a high school or college pitcher with two or three pitches. Adapt the numbers to your workload plan.
- Pitches 1 to 10: blocked fastballs to the glove-side half, from the stretch.
- Pitches 11 to 20: catcher calls fastball location late, alternating halves at random.
- Pitches 21 to 30: random pitch type and location, starting every hitter at 1-1, with a shot clock slightly shorter than your league's.
- Score it: a point for each pitch within a glove's width of the call, and track the total from week to week.
Be patient with the scoreboard. Driveline says command skill gains usually take a block of several weeks to show, even when strategy changes such as aiming adjustments help sooner [7]. Command takes a large sample to read, which Part 4 covers in detail.
Where to go next
Task constraints work best when they bring game information into practice. Part 3 covers exactly that: how to make bullpens more like real games. Part 4 shows how to progress a new pitch from the bullpen to the game.
Frequently asked questions
What are task constraints in pitching?
Task constraints are the rules, goals and equipment that shape a throw, such as target size, who calls the location, pitch order, the count, a pitch clock and ball weight. Coaches can change them quickly to make practice easier or harder.
Is random practice better than blocked practice for pitchers?
Blocked practice usually looks better in the moment, but random practice tends to transfer better. In a college hitting study, random pitch-type practice improved transfer-test performance by 56.7% versus 24.8% for blocked practice. Direct pitching trials are still limited.
Do weighted balls improve command?
Driveline uses 4, 5 and 6 ounce balls and slightly larger and smaller balls in its command programming, but there is no published outcome data yet showing improved pitching command. Treat it as a promising tool that still needs proof.
How hard should a command drill be?
Hard enough that you miss some of the time but succeed often enough to learn. Shrink the target or add a count when you start hitting it consistently, and make it easier if success collapses.
Command with a plan
Put constraint work on the calendar
Pitch AI builds dedicated Command days into your throwing program, and Pitch Lab logs every grip change alongside velo, IVB, spin, axis and release, so you can see what each change did to the shape.
References
- Boddy K, Daniels M. Hacking the Kinetic Chain: Advanced Pitching. Driveline Baseball; 2017. Chapter 3.1, Throwing Programming.
- Gray R. How We Learn to Move: A Revolution in the Way We Coach & Practice Sports Skills. 2021. Chapters 4 and 7.
- Driveline Baseball. Differential Command Program (program notes: command training balls about 5% larger and smaller, 4, 5 and 6 oz).
- Tread Athletics. "Brian Bannister | Mastering Pitch Design." YouTube, 2022-12-22. youtu.be/E0yoPt42A5Q
- Guadagnoli MA, Lee TD. Challenge point: a framework for conceptualizing the effects of various practice conditions in motor learning. Journal of Motor Behavior. 2004;36(2):212-224. doi:10.3200/JMBR.36.2.212-224
- Hall KG, Domingues DA, Cavazos R. Contextual interference effects with skilled baseball players. Perceptual and Motor Skills. 1994;78(3):835-841. doi:10.1177/003151259407800331
- Driveline Baseball. "The 2 Skills That Actually Make Up Pitching Command." YouTube, 2026-09-19. youtu.be/dXr7jj9ai4A
- 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.
- KNCT Athletics. "Fixing A Pushy Arm Action With Constraints." YouTube, 2026-09-22. youtu.be/J00_uPqEXis
- Rubicon. "Casting The Die | Episode 2 | Steffen Simmons." YouTube, 2026-09-19. youtu.be/PkRCz3dhYn0