Bar Speed (Velocity-Based Training)
Also known as: VBT, Velocity-Based Training, Bar Velocity, Mean Concentric Velocity, MCV
The concentric-phase speed at which a barbell moves during a lift, measured in metres per second. Bar speed is the objective input that velocity-based training (VBT) is built on: a given load produces a measurable velocity range in a given athlete, that velocity-load relationship is highly individual but very stable week-to-week, and changes in bar speed at familiar loads are one of the cleanest signals of fatigue or fitness shifts available outside a lab. VBT uses bar speed to autoregulate load (instead of percentages), to detect set-failure proximity (instead of RPE), and to corroborate the CNS-fatigue picture that subjective effort can't.
Formula
There is no formula — bar speed is measured directly by a velocity sensor (Vmaxpro, Enode, Beast, GymAware) attached to the bar, or computed from a phone-camera tracker. The standard metric is mean concentric velocity (MCV) in m/s. The athlete-specific load-velocity profile is built by lifting calibration loads (e.g. 30%, 50%, 70%, 90% of estimated 1RM) and recording MCV at each:
Load (% 1RM) Typical MCV (back squat) Typical MCV (bench press)
~50% ~1.10 m/s ~0.85 m/s
~70% ~0.80 m/s ~0.55 m/s
~85% ~0.55 m/s ~0.30 m/s
~95% ~0.35 m/s ~0.20 m/s
100% (1RM) ~0.30 m/s ~0.15 m/s
Minimum-velocity threshold (MVT) — the lowest MCV at which a successful rep can be completed — is athlete- and exercise-specific and remarkably stable (±0.05 m/s) over months.Example
Powerlifter builds a back-squat load-velocity profile in week 1 of a strength block. At 140 kg the first rep moves at 0.55 m/s; the prescribed working set's velocity loss stop-rule is 20% — so the set ends when bar speed drops to ~0.44 m/s. Two weeks later, the same 140 kg first-rep moves at 0.62 m/s: the load-velocity curve has shifted left, the athlete's 1RM has likely risen, and the next session can prescribe 145 kg targeting the same 0.55 m/s first-rep speed. No 1RM test required. Mid-block, on a poorly-slept morning, the first rep at 140 kg moves at 0.48 m/s: bar speed has corroborated the subjective 'something is off' read, and the prescription drops to 130 kg for that session rather than grinding through the planned load.
How Afitpilot Uses This
Afitpilot does not ingest bar-velocity data from sensors today. Bar speed enters the model implicitly via RPE — a sluggish set rates higher even at unchanged load, and the e1RM trend captures the load-velocity profile drift indirectly. VBT ingestion is on the same roadmap candidate as the CNS-fatigue and sticking-point surfaces would benefit from: optional integration with Vmaxpro / Enode / GymAware / Beast would let us overlay an objective velocity band on session cards, autoregulate load against a target first-rep velocity instead of (or alongside) RPE, and apply velocity-loss stop-rules as a third dimension of effort capture beyond RPE and RIR. Practical translation today: if you own a velocity sensor, the data tells you things RPE alone misses (bar speed dropping while RPE stays flat is a cleaner CNS-fatigue signal than either alone). If you don't, the existing RPE-and-e1RM stack covers most of the same ground at lower precision.
Bar speed in practice
| Who / Context | Value | Note |
|---|---|---|
| MVT (minimum-velocity threshold) — back squat 1RM | ~0.30 m/s (range 0.25-0.35) | Stable per athlete within ±0.05 m/s over months |
| MVT — bench press 1RM | ~0.15 m/s (range 0.10-0.20) | Lower than squat or deadlift because the press has less rebound assistance |
| MVT — deadlift 1RM | ~0.15-0.20 m/s | From the floor; deficit / above-knee variants shift the threshold |
| Typical velocity-loss stop-rule for strength | 10-20% loss vs. first-rep speed | Ends the set before the fatigue-vs-strength tradeoff turns unfavourable |
| Typical velocity-loss stop-rule for hypertrophy | 30-40% loss vs. first-rep speed | Allows more mechanical-tension accumulation per set at the cost of more peripheral fatigue |
| Where VBT outperforms %1RM prescription | Day-to-day readiness variation (5-15% load adjustment) | Same target velocity, different load on a poor-sleep day |
| Where VBT outperforms RPE prescription | Trained athletes; near-maximal singles | RPE 9 vs. RPE 10 is subjectively fuzzy; 0.30 m/s vs. 0.25 m/s is not |
| Calibration-load count for a useful profile | 4-6 loads spanning ~40-90% 1RM | Below 4 loads the curve fit is unreliable; above 6 the recovery cost outweighs the precision gain |
Known Limitations
- •Velocity sensors vary in accuracy. Linear position transducers (GymAware) are the lab gold standard at ±0.02 m/s; consumer accelerometer-based sensors (Vmaxpro, Beast) sit at ±0.05-0.08 m/s, which is meaningful relative to the target velocities (a 0.05 m/s error at 0.30 m/s 1RM speed is roughly 15% — enough to change whether a target rep is hit). Phone-camera trackers are noisier still; treat them as approximate.
- •Load-velocity profiles are exercise-specific and stance-specific. A profile built on competition-grip bench does not transfer to close-grip or paused-bench variants; rebuilds are needed when the prescription changes the movement meaningfully.
- •Bar speed degrades within a set faster than rep count or RPE suggest. The first 1-2 reps of a hard set carry most of the load-velocity signal; once velocity drops by 30-40% versus the first rep, additional reps tell you about fatigue tolerance, not strength quality.
- •VBT works best for compound barbell lifts with stable bar paths. Olympic lifts have phase-specific velocity demands that aren't captured by a single MCV; dumbbell, machine, and bodyweight work have neither the bar nor the path consistency the technology assumes.
- •Athlete calibration burden is real. A useful load-velocity profile requires 4-6 calibration loads with multiple reps each, repeated every 4-8 weeks as fitness shifts the curve. Athletes who set up the sensor once and never recalibrate are training against stale data within a mesocycle.
- •VBT's strongest practical use case is single-discipline strength sports (powerlifting, weightlifting) where the competition lift is the daily focus. Hybrid and general-fitness athletes get diminishing returns past the basic 'is bar speed dropping at familiar loads' read.
Science Context
Velocity-based training as a coaching modality traces to the work of Bryan Mann (Velocity-Based Training, 2016) and the Spanish school of González-Badillo, with the Sánchez-Medina & González-Badillo 2011 paper establishing the velocity-loss stop-rule as the operational core. The load-velocity relationship's linearity and within-athlete stability has been replicated across the squat (Jovanović & Flanagan 2014), bench press (García-Ramos et al. 2018), and deadlift (Banyard et al. 2017). The minimum-velocity threshold concept — that each athlete has a near-fixed MCV at which true 1RM occurs — is the mechanism behind VBT's load-prediction accuracy without max testing, with reported ±5 kg precision on 1RM estimates in trained populations (Pareja-Blanco et al. 2017). The honest scope limit for self-coached training: VBT's evidence base is strong but its practical advantage over RPE/e1RM-based autoregulation is largest in single-discipline strength athletes who lift the same competition movements weekly. For hybrid, general-fitness, and endurance-leaning athletes, the cost of sensor ownership and calibration usually outweighs the precision gain over the existing subjective stack. Afitpilot's roadmap candidate reflects this: surface VBT as an optional overlay for the athletes whose programming benefits most from it, not as a default everyone has to opt out of.