A wearable hands you a number — barbell velocity, muscle onset, jump height — with reassuring decimal places. Four 2026 validation studies ask how far to trust it, and land on the same shape of answer: reliable for a group or a trend, shakier for a single session, muscle, or device comparison.
Table of Contents
Reliable, up to a point
The good news is real: at the group level, agreement is often strong, with reliability coefficients sitting comfortably above 0.9. The catch is what sits next to them — individual accuracy is weaker, wide enough that one athlete’s single-session number carries real uncertainty, and devices are not interchangeable. A number can be reliable in aggregate and still mislead you about one person on one day.
Paper 01 — Do velocity sensors agree with each other?
A meta-analysis of 63 device studies and 38 prediction-model studies found commercial velocity sensors reliable at about 0.90–0.92, with linear transducers steadier than inertial units. But heterogeneity was large, especially in lower-body lifts — the pooled reliability is trustworthy while the spread across devices and exercises stays wide.
Paper 02 — Is muscle timing stable week to week?
Fifteen competitive rowers, seven muscles, two 2000 m trials a week apart. Activation onset was highly repeatable (reliability 0.94–0.99), but offset and peak position were more variable. Even within one athlete and one task, not every feature of the EMG timing is equally stable — which one you report matters.
Paper 03 — A wireless EMG sensor, muscle by muscle
Thirty-one resistance-trained participants, a wireless MR EMG sensor against a criterion EMG system, across several leg-extension loads. Within-session reliability was strong (0.92–0.97) — but it varied muscle by muscle, with the medial quadriceps reading more consistently than the lateral. Reliability is not a single badge for a device; it is per-muscle.
Paper 04 — Three jump tools, three jump heights
Twenty professional male volleyball players jumped while a force plate, the MyJump Lab 3 app and a Baiobit sensor measured simultaneously, retested 12 days later. The tools disagreed: the app overestimated jump height by 1.9–4.2 cm, and the wearable showed wide limits of agreement against the plate. Three tools, three heights — good for tracking change within one tool, risky for comparing across them.
Trust the trend, not the digit
The through-line across all four: reliability above 0.9 sits right next to individual errors of several percent. Linear transducers beat inertial units; medial EMG beats lateral; three jump devices give three heights. Wearables are excellent for tracking a trend within one device and protocol, and unreliable as an absolute cross-device truth. Trust the trend, not the digit. For the same lesson on the modelling side, see from kinematics to a modelled injury number.
Frequently asked questions
Are velocity-based training sensors accurate?
At the group level, yes — a 2026 meta-analysis put commercial sensors around 0.90–0.92 reliability, with linear transducers steadier than inertial units. But heterogeneity is large, especially in lower-body lifts, so individual-session numbers and cross-device comparisons carry more uncertainty than the headline coefficient suggests.
Is jump height from a phone app trustworthy?
For tracking change within that one app, largely yes; as an absolute number, be careful. Measured against a force plate, a MyJump-style app overestimated jump height by 1.9–4.2 cm, and a wearable showed wide limits of agreement. Three tools gave three heights — do not mix them.
Is EMG timing repeatable between sessions?
Partly. In competitive rowers, muscle activation onset was highly repeatable week to week (0.94–0.99), but offset and peak position were more variable — and within-session reliability also differed muscle by muscle. Which timing feature (and which muscle) you report changes how stable your numbers are.
References
[1] Claassen, N., Siegel, S. D., Sproll, M., et al. (2026). Reliability, Device Agreement and Validity of Load-Velocity Profiles: A Systematic Review with Meta-analysis. Sports Medicine – Open, 12, 102.
[2] Kresevic, S., Vignandel, E., Martini, M., et al. (2026). How Stable Are Temporal EMG Parameters in Rowing? A Seven-Day Test-Retest Reliability Study Using Wearable sEMG. Sensors, 26(15), 4914.
[3] Ghigiarelli, J. J., Gonzalez, A. M., Valdez, G. A., et al. (2026). Concurrent Validity and Within-Session Reliability of a Wireless Surface Electromyography Device (MR EMG) Compared to a Criterion Measure During the Leg Extension Exercise. Sensors, 26(16), 5165.
[4] Wilczynski, B., Sroda, T., Sikorski, M., et al. (2026). Agreement and reliability of force-plate, smartphone-based, and wearable-sensor countermovement jump assessments for neuromuscular monitoring in male professional volleyball players. Frontiers in Physiology.
Takashi Fukushima — Sports Science & Pose Estimation.
▶ Subscribe on YouTube · Website · ORCID · Contact