SSMT-2026-0010v12026-07-26Type M · MeasurementBiomechanicsdraft
Claim, as marketed
“Know exactly how fast you lifted.”
Canonical claim
Commercial velocity-based training devices measure barbell velocity accurately.
Scope
Linear position/velocity transducers and camera-optical devices only. IMU and accelerometer devices fail badly and are Contradicted at this claim. The verdict binds to the technology class, not the product category.
Verdict
SupportedSupported for linear transducers and camera devices. Contradicted for IMU/accelerometer devices.
This claim is split. The measurement limb. The practice limb — whether prescribing by velocity beats percentage-based prescription — is graded separately.
Magnitude drifts because 'commercial VBT devices' bundles a class where error reaches CV 75.8% with one where it is ±0.02 m/s.
Evidence located
6
primary studies
2
systematic reviews (44 & 75 studies)
0
vendor-authored studies found
17
largest n
Findings
Linear transducers. T-Force CV ≤0.62% with inter-device ICC ≥0.999; GymAware ICC 0.96–1.00, CV 2.5–7.7%, limits of agreement around ±0.03 m/s with no proportional bias. Criterion-grade for mean velocity.
Camera and optical. Mean-velocity bias ≤0.01 m/s and typical error of estimate 0.02–0.08 m/s. Accurate for mean velocity — though one device underestimates peak velocity by up to 0.12 m/s in the squat.
Smartphone apps are contested. One lab found CV 3.7–13.8% with no bias; another found smallest detectable change ≥0.24 m/s and recommended against use. Frame-rate and operator dependent.
IMU and accelerometer devices fail. Correlations of 0.33–0.86, CV rising to 38.3% at 100% 1RM, and in one head-to-head study CV of 75.8% with typical error of 0.48 m/s. Another device reached CV 60.5%.
The scale check that makes this actionable. Coaches autoregulate on 0.05–0.10 m/s bands. Transducer error of about 0.02 m/s sits comfortably below that. IMU error of 0.15–0.48 m/s is two to ten times the size of the decision being made.
Why reproducibility drifts. Of 56 reliability investigations in the largest review, only 2 separated technological from biological reliability, and only 5 of 66 validity investigations met all a-priori criteria. All validation is controlled-lab; no study validates devices under real training conditions — fatigue, sweat, chalk, off-axis bar path. One independent review explicitly warns against using these devices for force or power outputs.
What would change this verdict
Validation studies that separate technological from biological reliability, plus field validation under real training conditions. Downward pressure would come from evidence that consumer transducers or cameras drift in a crowded gym versus the controlled lab conditions of all current work.
Practitioner read
Buy the technology, not the promise. A linear position transducer or a camera unit will tell you bar speed to about ±0.02 m/s — accurate enough to autoregulate on. An IMU puck can be off by 0.15–0.48 m/s, which is larger than the decision you are making, so do not prescribe from one. Any metric based on peak velocity or derived power should be treated as unsupported.
Could not be verified
Funding not located for two studies — including the two device families that passed the strictest criteria. This is the most important unresolved item for the Independence score on this card.
Listed rather than dropped. Counts are floors, not censuses.
Sources
Thompson SW, et al. (2020). Sports 8(7):94 — 3D mocap criterion ↗
Weakley J, et al. (2021). Sports Med 51:443–502 — 44 studies, 22 devices ↗
Martínez-Cava A, et al. (2020). PLOS ONE 15(6):e0232465 ↗
Wannouch YJ, et al. (2025). PLOS ONE 20:e0324606 — 75 studies, 40 devices ↗
Every source must resolve at a DOI or PubMed ID.
Changelog
v1 · 2026-07-26 · Initial grade; first scope-qualified verdict in the register.
Disagree with this grade? Every card carries “what would change this verdict” so
the specification is explicit. Send evidence and it will be logged and the card
re-versioned in public — including when it moves a grade in a manufacturer's favour.
Submit evidence →
