Research theme
The first public presentation of the PhD’s whole-body optimisation results. The argument in one line: experimental research can tell you how bowlers differ, but it cannot tell an individual bowler what to change — for that you need a forward-dynamics model of that specific bowler.
Method, plainly: a 16-segment planar torque-driven computer simulation model of the front foot contact phase (front foot landing → ball release), built in AUTOLEV, with wobbling masses in the shanks, thighs and torso, nine torque generators (front MTP, front ankle, front knee, both hips, both shoulders, bowling elbow, bowling wrist) and the remaining joints angle-driven. Non-planar pelvis and shoulder rotation are approximated with two massless segments whose orientation and length are driven as functions of trunk angle; trunk side-flexion is handled by varying trunk length. Customised to one bowler using 18-camera Vicon motion capture plus a Kistler force plate at the ECB National Cricket Performance Centre, isovelocity-dynamometer strength measurements, and Yeadon’s inertia model. Evaluated, then optimised three ways with a genetic algorithm.
What they measured
- How fast the ball leaves the hand (ball release speed).
- How bent the front knee is through the phase (front leg kinematics).
- How far the bowler folds forward over the front leg (trunk flexion).
- When the bowling arm starts coming over (bowling shoulder extension timing).
- When the front arm starts coming down (non-bowling shoulder extension timing).
- The pose at the instant the front foot lands (initial body configuration parameters).
- How strong each joint is (subject-specific maximal voluntary torque–angle, torque–angular-velocity and differential-activation relationships from a nine-parameter function).
Findings
All causal within the model, single bowler. Numbers are rounded relative to the thesis (which gives 9.8%, 21.5%, 1.3%).
Model evaluation: 4% difference between simulation and recorded performance — good enough to optimise from.
Optimising movement only (same landing pose): +10%. Achieved by maintaining a straighter front leg and increasing trunk flexion.
Optimising the landing pose as well: +22%. “The most marked difference was at the shoulders where the extension was delayed for both the bowling and non-bowling arms.” That initial position then allowed the front leg to stay straighter and more trunk flexion to occur. Note the causal direction the paper asserts: the landing pose is upstream of the front-leg and trunk behaviour, not parallel to it.
Increasing strength by 5%: +1%. Maximum isometric torques at the ankle, knee, hip and front shoulder were raised 5%, with the optimal initial configuration held fixed. The gain was 1%, arising from “a straighter front leg, delayed trunk flexion and more extension of the front arm”.
The optimal technique was the same in all three optimisations: front leg kept straighter, bowling arm delayed, trunk flexion increased. The optimal body configuration additionally indicated the front arm’s extension should also be delayed, “most likely to aid in the delay of the bowling arm”. The authors note this agrees with what current elite fast bowlers do and with Worthington et al. (2013).
Stated ambition: to use the model “to directly support the coaching of elite fast bowling” — this cluster is explicitly a coaching-tool programme, not pure science.
What a coach should look for on video
This abstract supports four cues, all in the front foot contact phase. They are the same cues the thesis supports; see Felton 2015 — PhD thesis: factors limiting fast bowling for the underlying joint-angle numbers, which this 2-page abstract does not print.
Cue 1 — Both arms still “loaded” at the instant the front foot lands
- The cue: Where the bowling arm and the front arm are in the frame the front foot touches down.
- Camera view + frame: Side-on, at front foot contact.
- What “good” looks like: Bowling arm still back (extension not yet begun); front arm still high and not yet pulled down. This paper’s distinctive contribution is that both arms should be delayed, not just the bowling arm.
- What the fault looks like: One or both arms already unwinding at landing — the front arm dropped, the bowling arm halfway over.
- Why it matters: This was the single most marked difference in the +22% optimisation, and the paper argues delaying the front arm is what makes delaying the bowling arm possible.
Cue 2 — Straight front leg held through the phase
- The cue: Front knee behaviour from landing to release.
- Camera view + frame: Side-on, scrub front foot contact → ball release.
- What “good” looks like: Knee straight or straightening, not collapsing.
- What the fault looks like: Knee flexing after landing; head and hips sinking.
- Why it matters: Present in every one of the three optimisations. It is the most robust finding in this cluster.
Cue 3 — Trunk flexion increased
- The cue: How far the bowler folds forward over the braced front leg.
- Camera view + frame: Side-on, front foot contact → ball release.
- What “good” looks like: More fold than the bowler’s habit.
- What the fault looks like: Staying upright through release, or folding sideways rather than forward.
- Why it matters: Common to all three optimisations. Note the model treats it as an outcome of the straight front leg and the delayed arm, not something to be forced independently.
Cue 4 — Front arm pulled down actively (after the delay)
- The cue: Once the front arm does start, does it come down hard and into the torso?
- Camera view + frame: Front-on or side-on, from front foot contact to release.
- What “good” looks like: Delayed start, then a decisive pull-down. The strength optimisation specifically produced “more extension of the front arm”.
- What the fault looks like: A passive or sideways-drifting front arm.
- Why it matters: Appears in both the initial-configuration and strength optimisations.
Not supported: any cue about strength training. A 5% strength increase bought 1%.
Caveats and limits
- n = 1. A single male fast bowler: age 18 years, mass 85.0 kg, height 1.935 m, England U19 squad, identified as a potential England player within five years. This is the same bowler as the thesis and the 2017 and 2020 papers.
- Two-page conference abstract. Almost no numerical detail: no joint angles, no ground reaction forces, no per-component match scores, no confidence bounds. For the underlying numbers, use the thesis or the 2020 journal paper.
- Simulation, not intervention. Nothing tested on a real bowler.
- Planar (2D) model. Non-planar rotation approximated.
- Front foot contact phase only. Cannot say whether the bowler could physically reach the optimal landing pose, or what it would cost in the run-up and back foot contact phase.
- Individual-specific by design. The percentages belong to this bowler.
Relationship to other Felton work
- This and Felton 2017 — optimising individual performance report the same three optimisations on the same bowler — the 2017 ISBS paper is a longer, more methodologically detailed write-up of this same content, with the same 10% / 22% / 1% numbers.
- Both are conference presentations of PhD Chapter 9 (Felton 2015 — PhD thesis: factors limiting fast bowling), which gives the unrounded 9.8%, 21.5% and 1.3%.
- The peer-reviewed journal version is Felton 2020 — optimising the front foot contact phase.
CONTRADICTION: (conference vs journal, on the size of the prize) This abstract’s headline claim is that optimising the landing position is worth 22%. The peer-reviewed 2020 J Sports Sci paper on the same bowler and the same model reports only the 9.8% movement-only optimisation and states that varying the initial position of the bowling arm “was outside the scope of this paper”. The 22% figure never appears in the journal literature for this bowler. When the equivalent both-position-and-movement optimisation was finally published for a group (2023 J Biomechanics, ten bowlers), the mean gain was 13.5%, not 22%. Coaches quoting “22% from a better landing position” are quoting a conference abstract that was never replicated at that magnitude.
TENSION: (on how much strength matters) this abstract reports that increasing strength 5% produced a straighter front leg, delayed trunk flexion and more front-arm extension — i.e. strength reinforced the optimal technique. The 2025 J Sports Sci paper, running the same manipulation across ten elite bowlers, found the non-significant trend ran the opposite way (less knee extension, reduced trunk flexion) and called it “contrary to expectations”. See Felton 2025 — the effect of increased strength on ball release speed.