2022 — Pitch Length Variations and the Ability to Control Length

Research theme

Almost all fast bowling biomechanics up to this point had treated ball release speed as the performance outcome. In T20-era cricket, where the ball lands matters at least as much. This study asked two things: (1) do the ball’s release conditions actually differ between a yorker, a stock ball and a bouncer, and (2) how reliably can good bowlers hit the length they intend? Method: 18-camera Vicon motion capture at 250 Hz of 21 male county academy fast bowlers (height 1.87 ± 0.05 m, mass 81.40 ± 9.74 kg), indoors on an artificial surface, 36 deliveries each in randomised order; a synchronised 2D camera behind the run-up plus 2D direct linear transformation measured where each ball actually pitched. 707 deliveries analysed. This is a descriptive measurement study — everything below is correlational/observational, not causal.

What they measured

• Where the ball actually landed, in metres from the batter’s stumps (pitch length, via 2D DLT against chalk lines every 2 m out to 12 m).
• Delivery target zones: yorker 0–2 m, stock 4–7 m, bouncer >7 m from the batter’s stumps.
• How fast the ball left the hand (resultant ball release speed, m/s).
• The upward tilt of the ball’s flight path the instant it leaves the hand (ball release angle, degrees above horizontal, in the sagittal plane).
• How high the ball was released, expressed as a percentage of the bowler’s own standing height (so tall and short bowlers are comparable).
• Whether each ball landed in its intended zone (success rate).

Findings

1. Bowlers hit a bouncer almost every time and a yorker almost never. Success rates: bouncer 98.7% (233/236), stock 46.4% (110/237), yorker 24.8% (58/234).
2. The mean yorker was not a yorker. Mean landing point: yorker 3.8 ± 3.3 m (target 0–2 m, so the average ball was ~1.8 m too short of the longest permitted yorker); stock 7.0 ± 3.7 m (exactly on the shortest edge of its 4–7 m band); bouncer 10.8 ± 1.4 m (comfortably inside). One-way repeated-measures ANOVA F(2,704) = 499.5, p < 0.05; all three lengths differed from each other on post-hoc Tukey (p < 0.05).
3. The spread, not just the mean, is the story. Standard deviation of landing point was 1.4 m for bouncers versus 3.3 m (yorker) and 3.7 m (stock) — roughly 2.5× more scatter on the fuller lengths, despite the bouncer having by far the biggest legal target area.
4. Release angle is the parameter that separates lengths, and it separates them cleanly. Yorker 1.9 ± 1.8°, stock 5.3 ± 1.4°, bouncer 13.0 ± 2.7°. Main effect p < 0.001 with a very large effect size η² = 0.79, and it is the only release parameter significantly different across all three pairwise comparisons (all p < 0.001).
5. Release speed barely moves. Yorker 31.5 ± 2.3 m/s, stock 31.8 ± 3.3 m/s, bouncer 33.1 ± 2.2 m/s. Main effect p < 0.001 but a small effect size η² = 0.07; post-hoc, bouncers were faster than both others (p < 0.001) while stock vs yorker did not differ (p = 0.710).
6. Release height moves a moderate amount, and in the opposite direction to intuition. As a percentage of standing height: yorker 112.2 ± 4.1%, stock 110.7 ± 7.9%, bouncer 107.4 ± 4.0% (η² = 0.22). The bouncer is released lowest — the arm has come further forward and down. Bouncer differed from both (p < 0.001); stock vs yorker did not (p = 0.173).
7. The margin for error narrows dramatically as you pitch fuller. The range of release angles that still produced a successful ball was 16° for bouncers, 9° for stock and only 5° for yorkers. That, more than any technique fault, is why the yorker misses.
8. Practical conclusion drawn by the authors: ball release angle can be used as a surrogate for pitch length in future biomechanics work — i.e. if you can change release angle, you can change length.

What a coach should look for on video

The paper measures the ball, not the body, so the honest coaching cues here are about release conditions and about which errors to stop chasing. (The body-side cues arrive in the 2023 paper and the thesis — see the sibling files in this folder.)

• Cue: the initial upward tilt of the ball out of the hand.

  • Camera view + frame: Side-on, camera level with the crease, phone in slow-motion. Scrub to ball release and step 3–5 frames forward. Draw the line the ball travels against the horizontal.
  • What “good” looks like: roughly 2° for a yorker, 5° for a good length, 13° for a bouncer. In practice: a yorker should look almost flat out of the hand; a bouncer visibly climbing.
  • What the fault looks like: a bowler who is “trying to bowl a yorker” but whose ball leaves the hand at good-length tilt (~5°). Only about 3° separates a perfect yorker from a half-volley at the batter’s feet, so this is a tiny, real difference you have to look for deliberately.
  • Why it matters: release angle explained 79% of the between-delivery variance. It is the lever.

• Cue: stop coaching “bowl it faster/slower to change length.”

  • Camera view + frame: any — this is a radar/speed-gun cue, not a video one.
  • What “good” looks like: release speed essentially unchanged between yorker and stock (31.5 vs 31.8 m/s, p = 0.71). Bouncers came out only ~1.3–1.6 m/s (≈5 km/h) faster.
  • What the fault looks like: a bowler slowing down to “get it full”. The data say speed is not how these bowlers changed length, and taking pace off is pure cost.
  • Why it matters: changing length via speed sacrifices the thing that makes the ball hard to hit.

• Cue: for a bouncer, expect the release point to be LOWER, not higher.

  • Camera view + frame: Side-on, freeze at ball release. Compare ball height to the top of the bowler’s head across a bouncer and a yorker from the same session.
  • What “good” looks like: bouncer released at ~107% of standing height, yorker at ~112% — the bouncer roughly 5% of body height lower, about 9–10 cm on a 1.87 m bowler.
  • What the fault looks like: a bowler reaching up and trying to “bang it in from height”. In this sample the short ball came from the arm having circumducted further forward and down.
  • Why it matters: it explains why the bouncer is also the fastest ball — more time and range of motion to accelerate the hand through.

• Cue: judge yorker practice on spread, not on the best ball.

  • Camera view + frame: Behind the arm / high overhead, mark the pitch in 2 m bands.
  • What “good” looks like: the bouncer’s 1.4 m SD is what “controlled” looks like. County academy bowlers ran 3.3 m SD on yorkers — three of four missed.
  • What the fault looks like: a session judged by the two yorkers that nailed it.
  • Why it matters: the paper’s own framing — deliveries that must travel further have smaller windows of successful release conditions, so measure the window.

Cue this paper does NOT support: anything about the front leg, the run-up, the delivery stride or the action type. None of those were measured here.

Caveats and limits

• n = 21, all male, all county academy (young senior/junior elite) — not adult internationals, not women, not club bowlers.
• Indoor artificial surface with no batter present. The authors flag this explicitly: with no physical target and no match pressure the task is visually and psychologically different, and yorker accuracy in particular could go either way in a match.
• The yorker success rate is arguably harsh: a 2 m-wide target versus a >7 m open-ended bouncer zone. Some of the 98.7% vs 24.8% gap is task geometry, not skill — the authors say so.
• Purely correlational/descriptive. Nothing here shows that changing a technique parameter causes a length change; the technique link comes later.
• Only successful trials (401 of 707) went into the release-parameter comparison, so the release numbers describe well-executed deliveries, not the misses.

Relationship to other Felton work

• This is the first output of Kaushal Manawadu’s PhD, supervised by King, Hiley, Felton and McErlain-Naylor. It sets up the entire sub-line: it establishes release angle as the outcome variable, and the follow-up work goes hunting for the technique that produces it.
• Directly extended by Manawadu et al. 2023, which regresses release angle onto body kinematics, and by Manawadu 2023 PhD thesis, which contains this study as Chapter 5 with extra per-bowler data and a projectile-motion sensitivity model.
• It deliberately steps away from the Felton/King mainline (Worthington et al. 2013; Felton, Yeadon & King 2020) which optimises for ball release speed. This paper’s finding that front-of-body technique changes length while speed stays flat is the beginning of a second performance axis.

No contradiction with other Felton work. One internal inconsistency worth noting: the abstract says release speed and height “of the bouncer deliveries were significantly different to yorker and stock”, which is correct, but the ANOVA table’s stock value for release height is printed as 110.7 ± 7.9 (99.0 ± 118.6) — the range is mistyped with a ± instead of a –. The thesis (Chapter 6) reports the same variable as 111.3 ± 3.9%, a slightly different figure, because the thesis analysis uses one representative trial per bowler rather than all successful trials.