How to check if your bowling action is legal from video

The self-check in ten steps

1. Set the phone to its highest frame rate: 120 or 240fps slow motion if it has it, and never below 60fps.
2. Put it on a tripod side-on to the crease on your bowling-arm side, roughly level with your release point, far enough back that the whole arm stays in frame from horizontal to release.
3. Film in good light, outdoors or in bright nets: high frame rates allow only short exposures, so dim light leaves the arm dark, noisy and hard to place.
4. Bowl in short sleeves or a sleeveless top, with a strip of contrasting tape on the tip of the shoulder, the outside of the elbow and the back of the wrist.
5. Bowl at match pace off your full run-up, and film at least six of each delivery you use — stock ball, bouncer, yorker and every variation.
6. If you can, add a second phone behind the bowler or at mid-off, to see whether your arm is swinging square to the first camera.
7. For each delivery, find the frame where the upper arm reaches horizontal and the first frame where the ball has left the hand.
8. Measure the shoulder–elbow–wrist angle in every frame between those two, and note the largest straightening from the most-bent reading.
9. Judge each delivery on its own and look hardest at the worst one: the ICC test fails an action if any single delivery exceeds 15 degrees.
10. If any delivery reads near or over the limit, or your two views disagree, stop self-diagnosing and arrange a formal assessment through your board or state association.

Filming it properly, and why each step matters

Frame rate decides how finely the window is sampled. At 30 frames per second consecutive frames are 33 milliseconds apart; at 240 they are about 4 milliseconds apart. The window from arm horizontal to release is a small fraction of a second, so a low frame rate gives a handful of samples and can miss the most-bent or straightest point. Formal procedures ask for far more: one English county board’s protocol for non-first-class cricket asks clubs for 200 Hz high-speed video, and ICC-accredited centres use at least 12 high-speed cameras.

Camera position matters because a camera only sees bending in a plane facing it. Research on why throws are hard to detect found that the plane in which the elbow bends rotates by about 90 degrees during the delivery swing, because the upper arm rotates, so elbow extension is seen undistorted only from exactly 90 degrees to that plane (Aginsky and Noakes, British Journal of Sports Medicine, 2010). No single position sees the whole window perfectly. Side-on to the bowling arm, level with release, is the view to start with; formal reviews use several, and the ICC’s assessment report records side-on, behind and front-on match footage.

Clothing changes the numbers. When a shirt covered the arm in field tests, the error in elbow angle grew for both methods that located the joints by eye (Elliott, Alderson and Denver, 2007). The ICC’s protocol tests men bare-chested with markers fixed to the arm and torso. Short sleeves and three strips of tape give you consistent points to click on in every frame.

Bowl the deliveries you actually use. The ICC test asks for 18 to 36 balls, at least six of any type that was reported, and splits a fast bowler’s stock deliveries between good length, yorker and bouncer. In match footage of elite fast bowlers, bouncers and short balls showed more straightening on average than good-length or full deliveries, although the difference was not statistically significant (Portus and colleagues, 2006).

Reading the elbow angle frame by frame

The free route is manual. Use video-analysis software that can show time as frame numbers, can be told the real capture rate of a slow-motion clip so timings are correct, and has an angle tool. Free programs that do all three exist, some of them open source; check that the one you choose runs on your computer. Find the frame where the upper arm passes horizontal, place the angle’s three points on shoulder, elbow and wrist, and record the reading. Then step forward one frame at a time to the first frame where the ball has clearly left the hand, recording the angle in each.

Elbow extension is the straightening inside that window: the largest drop from the most-bent reading to the straightest later one, ignoring any movement past fully straight, which the ICC discounts as hyperextension. Do not simply subtract the release angle from the angle at horizontal, because the elbow can keep bending after horizontal before it straightens. The ICC’s report template records the angle at horizontal, the maximum, the minimum and the angle at release, not just two numbers.

Doing this by hand for every frame of every delivery is slow, and every click carries some placement error. Automatic tracking removes the tedium but not the geometry. Stance measures elbow angle across the window from ordinary phone footage, over a session of several marked deliveries; it is information about your action, not an official ruling, and it is bound by the same single-camera limits described next.

What one camera can and cannot tell you

The argument that a phone cannot check legality is right about the verdict. Only an accredited assessment counts, and the official protocol controls the cameras, calibration and processing. It is also right that unchecked automatic numbers can be badly wrong: a 2025 study comparing a markerless pose-estimation system with marker-based capture in cricket bowling found an average elbow-angle error of about 23 degrees — more than the limit itself — and only 217 of 473 trials were usable in both systems (Abraham, Feros and Fox, 2025).

Projection is the core problem. One camera flattens a 3D movement, and a bend in a plane turned away from the lens looks smaller: in the simplest case by roughly the cosine of the turn, so about 3 per cent smaller at 15 degrees off, 13 per cent at 30 and half at 60. Because the elbow’s bending plane rotates during the swing, that error is not constant across the window, and from the wrong view a fixed carrying angle at the elbow can look like a bend. Placing joints by eye adds more: on a mechanical arm with known angles, the two methods that located joints visually had root-mean-square errors of 4.5 and 9.4 degrees (Elliott, Alderson and Denver, 2007).

What the argument overstates is that video carries no useful information. The ICC’s own system uses it: for international matches outside Tests, one-day internationals and T20 internationals, panels of human-movement specialists decide from video whether an action is illegal, and from November 2020 to the end of 2021 reported bowlers were assessed that way, with a lab test available on challenge where practical. Loughborough researchers have used elbow angles estimated from high-speed video as the reference for checking marker methods (Yeadon and King, 2015). Specialists reading video are not an app, but the information is in footage that is filmed well.

So bound the error rather than ignore it: keep the arm square to the main camera, confirm with a second view, use tape so joints are placed consistently, compare deliveries filmed the same way rather than trusting one absolute number, and treat any reading within several degrees of 15 as unresolved.

Interpreting what you get

Comfortably inside: every delivery, including bouncers, yorkers and variations, straightens well under 15 degrees with a margin larger than the measurement error, and the second view agrees. Film again from time to time; when the ICC introduced its 2005 process it stopped permanently clearing reported bowlers precisely because actions can change over a career.

Near the limit: any delivery within several degrees of 15, or readings that shift noticeably between camera views. This is the zone a single camera cannot settle in either direction. Work on the action with a coach, film again under the same set-up, and consider formal testing.

Worth formal testing: any delivery clearly over 15 degrees, or one delivery type consistently higher than the rest. The ICC assesses variations separately: if only a variation such as the doosra fails, the bowler may carry on with a warning, though a second report and failed test on it brings immediate suspension. Bouncers and yorkers count as the stock ball, not as variations. If your elbows are very flexible, say so: accredited centres screen for joint hypermobility on the Beighton scale and discount hyperextension, which a phone clip cannot do.

Does fatigue change the action across a spell?

Less than is often assumed, at least for fast bowlers tested in a lab. Junior fast bowlers bowling a 10-over spell at competition pace showed no substantial change in the mean or variability of any kinematic variable measured, which included the bowling arm (Schaefer and colleagues, 2018). Young pace bowlers over an eight-over spell likewise showed no significant change in trunk, knee and shoulder kinematics or ball speed (Callaghan and colleagues, 2019).

Neither study set out to measure elbow extension against the legality limit, and neither covers spinners or a long day in the field, so the cheap safeguard is to film late in a spell as well as fresh. What moves the reading more predictably is the delivery itself, which is why every length and variation you bowl needs its own clips.

Getting an official test

In international cricket an official assessment follows a match official’s report; the ICC arranges it at an accredited centre and pays for the first one, while re-assessments after a suspension are at the player’s or board’s cost. The ICC’s regulations list five centres: Lahore, Loughborough, Chennai, Brisbane and Pretoria.

In India the BCCI’s published league policies show the domestic route. A player on the warning list, or suspended, may request an Official Assessment from the BCCI Suspect Bowling Action Committee, but must first have a 3D biomechanical analysis at an ICC-approved centre; the BCCI will try to arrange it at the Sri Ramachandra Arthroscopy and Sports Sciences Centre in Chennai, and the player or franchise pays for the analysis. For state or age-group cricket, ask your state association how it handles suspect actions — the ICC requires every board to have a domestic procedure.

Take your best footage with you. A clear side-on clip of several deliveries of each type is what helps a coach or panel decide whether testing is needed, and the ICC’s own process starts from match video.