Source note: full accepted manuscript read, including Table 2 and all reported SPM cluster results. All numbers below are transcribed directly.

Research theme

Almost all cricket biomechanics research, and almost all winter training, happens indoors on artificial turf. Almost all cricket is played on natural grass. Nobody had checked whether the action is the same on both. This matters twice over: for coaches (does net footage transfer to the middle?) and for the entire research literature (are the artificial-surface studies that underpin coaching even valid?). Eight sub-elite male fast bowlers (University Centre of Cricketing Excellence, comparable to English County 2nd XI; age 19.4 ± 1.3 yr, height 1.83 ± 0.09 m, mass 77.4 ± 9.1 kg) bowled a minimum of six maximal good-length deliveries outdoors on a natural grass pitch on day one, then indoors on artificial turf the next day — an 11 mm pile monofilament polyethylene carpet over a 15 mm bound rubber shockpad over concrete, with room for a full run-up. Both sessions captured with an 18-camera Vicon system at 300 Hz, 47 markers, the same researcher placing markers both days. Analysis: paired t-tests on discrete variables plus statistical parametric mapping (SPM) on the whole time-normalised joint-angle waveforms, so a difference anywhere in the phase would be detected, not just at key instants. Exploratory; no hypothesis posed. Measured comparison, within-subject.

What they measured

Findings

The one significant difference: the feet slide more indoors

  1. Back foot contact slip distance: 7.1 ± 3.5 cm on grass vs 8.8 ± 3.6 cm on artificial; mean difference −1.7 ± 0.8 cm, p = 0.001, d = 2.12 (95% CI 0.81–3.39) — a very large effect.
  2. Front foot contact slip distance: 6.6 ± 2.5 cm on grass vs 8.7 ± 2.0 cm on artificial; mean difference −2.1 ± 1.8 cm, p = 0.013, d = 1.16 (95% CI 0.23–2.06) — large.
  3. Attributed to traction: non-spiked trainers on carpet grip less than 6 mm spikes in turf.

Everything else: no difference 4. Ball release velocity: 30.2 ± 2.2 m/s (grass) vs 30.4 ± 2.2 m/s (artificial), p = 0.438, d = 0.29. Bowlers were, if anything, a fraction quicker indoors. 5. Run-up velocity: 5.4 ± 0.4 vs 5.4 ± 0.3 m/s, p = 0.282, d = 0.41. 6. Back foot contact → ball release: 0.328 ± 0.032 vs 0.327 ± 0.038 s, p = 0.813, d = 0.09. 7. Back foot contact → front foot contact: 0.218 ± 0.033 vs 0.215 ± 0.034 s, p = 0.336. 8. Front foot contact → ball release: 0.110 ± 0.008 vs 0.111 ± 0.009 s, p = 0.331. 9. Whole-body COM acceleration BFC→FFC: −3.7 ± 0.6 vs −3.8 ± 0.7 m/s²; FFC→BR: −11.0 ± 2.5 vs −11.4 ± 1.8 m/s². Pelvic COM acceleration BFC→FFC: 1.2 ± 1.0 vs 1.0 ± 1.3; FFC→BR: −25.5 ± 4.3 vs −26.4 ± 4.2. All non-significant (p = 0.28–0.81 across the whole discrete set, d = 0.09–0.41). 10. SPM found no statistically significant kinematic difference anywhere, in any joint angle, at any point in either phase (p > 0.05). This is the headline. Not “no difference at the key instants” — no difference across the entire time-normalised waveform.

Exploratory sub-threshold trends (NOT statistically significant — reported by the authors where the 95% CI of the mean difference excluded zero) 11. Back foot contact phase, artificial vs grass: less front knee flexion (99–100% of phase, peak mean difference 4°); greater ipsilateral pelvic twist (73–100%, peak 7°); greater contralateral lumbopelvic side flexion (43–48%, peak 2°). 12. Front foot contact phase, artificial vs grass: greater rear knee flexion (0–21%, peak 3°); less rear hip flexion (50–87%, peak 2°); greater ipsilateral pelvic twist (0–49%, peak 7°); greater contralateral lumbopelvic side flexion (92–100%, peak 4°); less ipsilateral thoracolumbar rotation (85–100%, peak 2°). 13. The largest of these is the 7° pelvic twist, present in late back-foot and early front-foot phases. The authors speculate this is the adaptation that absorbs the extra slip and keeps the business end of the action (thoracolumbar flexion, shoulder extension at front foot contact) unchanged — i.e. the bowler reorganises upstream so that ball speed is preserved.

The injury-risk verdict 14. The two best predictors of prospective lumbar bone stress injury (Alway et al. 2021 — rear hip flexion at back foot contact, and lumbopelvic flexion/extension at front foot contact) showed no difference between surfaces. The one rear-hip trend occurred at 50–87% of the front foot contact phase, i.e. long after back foot contact, and the authors judge it “unlikely to influence lumbar bone stress injury risk.” 15. Lumbopelvic flexion, front hip flexion, pelvic tilt and thoracolumbar side flexion: no differences at all. Other injury-relevant variables (thoracolumbar rotation, lumbopelvic side flexion, rear knee flexion, pelvic twist) showed only sub-threshold trends, and at phase timings different from those previously linked to injury. 16. Practical consequence stated by the authors: artificial-surface deliveries “should be included as part of any workload monitoring” — they load the back the same way grass deliveries do.

What a coach should look for on video

The practical verdict a coach actually needs: on this evidence, net-session footage on artificial turf transfers to the middle. Everything you would coach off video — front knee, trunk flexion, arm position, hip/shoulder alignment, front foot contact timing — was statistically indistinguishable between surfaces, across the entire movement, in the same eight bowlers two days apart. Ball speed was also unchanged. You do not need to discount your indoor footage.

Cue 1 — Foot slide at landing (the one thing that genuinely changes indoors)

Cue 2 — Hip/pelvis alignment at back foot contact, if you are comparing indoor to outdoor footage of the same bowler

Cue 3 — Workload counting

Cues this paper explicitly does NOT support: any claim that artificial nets push a bowler towards or away from injury-risk positions, or that they cost ball speed. Both were tested and both came back null.

Caveats and limits

Relationship to other Felton work