What VBT measures

Every repetition has a velocity, so VBT gives an objective, immediate readout of output on every rep. It needs technology to capture it: a linear position transducer, an inertial sensor on the bar, a camera system or a smartphone app.

It helps to be precise about what VBT is. The Definitions and Diagnostics review argues that VBT is often described as a training method (“we’re doing VBT”) when it is better seen as an umbrella of guiding principles: using velocity to inform testing, monitoring, programming, autoregulation and feedback. Load-velocity profiles and velocity loss thresholds are the main tools under that umbrella.

Which velocity?

Most systems report one of three metrics, and they are not interchangeable. Thresholds set in one metric don’t apply to another.

Reliability is similar across the three. But peak velocity is a single snapshot, so it is more easily distorted by sampling rate, filtering, technique and measurement error. In weightlifting moves, mean velocity is dragged down by the slow first pull. For most coaches, the review recommends mean velocity as the default for every exercise.

The reason propulsive velocity exists is the braking phase. In a non-ballistic lift, the athlete has to slow the bar down at the top so they don’t leave the ground or let go of it. At light loads that braking takes up a big share of the lift: in the bench press, only about 76% of the concentric phase is propulsive at 30% 1RM, rising to about 86% at 50% and 100% from around 80% 1RM. That is why light-load velocities in squats and presses are less reliable, and why ballistic versions help (see below).

Why a fixed 1RM falls short

Percentage-based programming assumes the 1RM tested at the start of a block holds all the way through. It doesn’t. Strength rises and falls with fatigue, sleep, stress, nutrition and the phase of training, so the athlete’s real maximum on any day sits somewhere in a band around the plan.

Coaches have several ways to adjust for this, known as autoregulation: ratings of perceived exertion, repetitions in reserve, autoregulatory progressive resistance exercise (APRE) and flexible non-linear programming. All of them have produced better strength and power gains than fixed periodisation alone, but each has a weakness. Rating scales are subjective, APRE relies on training to failure, and flexible non-linear programming has little evidence behind it. Repetition maximums, meanwhile, push athletes to failure and can create excessive fatigue.

VBT offers a more objective option. Because velocity falls in a predictable way as load rises, the bar speed tells you where the athlete is today, without asking them to guess or grind to failure.

The load-velocity relationship

The physics is simple. Force equals mass times acceleration, and power equals force times velocity, so velocity is built into the mechanics of everything strength coaches train. At the muscle level, faster shortening leaves less time for the actin-myosin cross-bridges to form, so force falls as speed rises: the force-velocity relationship. In multi-joint lifts, this relationship becomes close to a straight line.

Plot several loads against their bar velocities and you get the athlete’s load-velocity profile for that lift. Across more than two decades of research, the relationship has been near-perfect (r and R² above 0.9) in every exercise studied, from the back squat, deadlift and bench press to the power clean, pull-up and jump squat. Day-to-day reliability is good too (CVs of 1.4–10%).

Three practical points follow:

• Profiles are individual. Athletes vary a lot in velocity at the same relative load, so individual profiles give more accurate estimates than group averages. That said, one 6-week study found no difference in performance gains between individual and group-based profiles (Dorrell et al., 2020), so normative data is a reasonable start if you are short of time.
• Profiles are specific to the lift and to sex. A squat profile doesn’t transfer to the bench press or trap bar deadlift, and men typically show higher velocities than women at the same relative load.
• Beware Smith machine data. Much of the research used Smith machine bench press. Free-weight velocities, loads and muscle activity differ, so don’t apply those numbers to free-weight lifts.

The cost is logistics: testing time, rest between efforts and periodic re-testing as athletes adapt. Barbell velocity also describes the bar, not the athlete’s centre of mass, so don’t read whole-body power from it. Barbell-based methods have been shown to overestimate jump power by 28–63% compared with force plates.

Where the zones typically sit

Published velocity zones for mean velocity in lower-body compound lifts:

Three qualifications matter:

• Zones depend on the metric. Bands built for mean velocity don’t apply to mean propulsive velocity, and the two differ most at light, fast loads.
• Zones depend on the lift. The deadlift and bench press boundaries differ from the squat, as the table above shows. At 1RM, the bench press moves at about 0.18 m/s and the deadlift at about 0.33 m/s.
• Zones depend on the athlete. Zones show you where to look; an individual profile shows where the athlete actually is.

Building a profile

A profile is built like a 1RM test: velocity is recorded across a range of loads, then a line (or curve) is fitted to the data. The Definitions and Diagnostics review sets out the decisions in order:

A note on the maths: load is the independent variable (x) and velocity the outcome (y). To predict load from a velocity, rearrange the equation to solve for x rather than swapping the axes. In Excel, =LINEST(velocity, load, TRUE, FALSE) gives the linear coefficients, and =LINEST(velocity, load^{1,2}, TRUE, FALSE) gives the quadratic ones.

Velocity loss: the strongest evidence in VBT

Instead of prescribing a number of reps, set a velocity loss threshold: a percentage drop from the fastest rep in the set. When a rep falls below it, the set ends. The rep count becomes an output, not an input.

The key study is Pareja-Blanco and colleagues (2017). Twenty-two trained men squatted twice a week for eight weeks at about 70–85% 1RM, three sets per session. The only difference between groups was the velocity loss allowed in each set: 20% or 40%.

The 20% group did 40% fewer reps, gained the same strength, jumped considerably better and kept their fast-twitch fibres. The caveats: the sample was small, there was no control group, and the extra muscle went to the higher-volume group. If hypertrophy is the goal of the block, a tight threshold is the wrong tool.

Weakley et al. (2021) explain why thresholds work so well. Each step up in velocity loss (10%, 20%, 30%) causes a roughly linear increase in neuromuscular fatigue, perceived effort and metabolic stress, and these responses are consistent within and between athletes. Because each athlete does as many reps as it takes to reach the same relative drop, a threshold creates a ‘flexible-repetition’ scheme that adapts to individual work capacity and readiness.

Choosing a velocity loss cut-off

The evidence supports a direction, generally smaller than most coaches pick, rather than one perfect number. A starting framework:

• 10–20% when speed, power and availability come first: in-season, or any block where the athlete has to be fast at the weekend.
• 20–25% when strength is the priority and fatigue needs to be controlled.
• 30–40% when hypertrophy and volume are the point of the block.

The 20% figure carries the most weight, because it is the one tested directly against a larger threshold. Two cautions: a fixed threshold can produce very different rep counts in the same squad (one study reported about 2 to 11 reps at a 10% loss), and a 40% threshold sent about 56% of sets to failure in another, which undermines the fatigue control you’re after. Pairing a velocity threshold with a rep cap keeps the intent of the session.

Weakley et al. (2021) also show how thresholds fit traditional periodisation: for example, 30% loss in a strength-endurance block, 20% in a strength block and 10% in a power or taper block.

A worked example

This example is illustrative, not a real athlete. The athlete back squats with a 20% velocity loss threshold. Their best recent velocity at this load is about 0.70 m/s, so the set ends when a rep drops below 0.56 m/s.

The set ends at six reps because the threshold was crossed, not because six were programmed. The next week, the same athlete crosses 20% at rep four. That is the method working as intended: four good reps were available that day, and the alternative was grinding out six and paying for it on Saturday.

One short set is noise. Three sessions of progressively earlier crossings at a known load is a signal: reduce the load, reduce the exposures, or look at what is happening outside the gym.

What VBT doesn’t tell you

Devices aren’t equivalent. Linear position transducers show better validity and lower error than bar-mounted accelerometers and inertial sensors, several of which are biased or only accurate at slow speeds (Weakley et al., 2021b). Device noise can be larger than the ±0.06 m/s change coaches often treat as meaningful.

A daily 1RM needs the right method. Predicting free-weight 1RM from a simple straight line is not accurate enough: studies report systematic bias of about 7–38 kg (6–20%), errors that grow in stronger athletes, and nearly 10 kg error for the two-point method in the free-weight bench press. The newer evidence is more encouraging. Using a quadratic model with at least four loads, and predicting from the velocity at 80% 1RM (the heaviest load with reliable velocity) rather than from a minimum velocity threshold, cut the error in the free-weight back squat to about 1 kg (Thompson et al., 2021a). That is one study, so direct testing remains the gold standard until it is replicated.

It doesn’t replace a profile. Velocity can track fatigue, and drops after a heavy session typically recover within about 72 hours. But interpreting small changes needs reliable data and an exercise-specific profile.

It can degrade technique. Athletes chase numbers and form drifts. That’s a coaching problem, not a technology problem.

It suits some phases better than others. VBT is most useful in pre-season and in-season. Tight thresholds work against high-volume off-season accumulation.

When to override the numbers

If the athlete looks wrong but the data looks fine, trust what you see. Velocity doesn’t measure soft-tissue irritation, sleep, illness or anything the athlete hasn’t told you.

If the athlete says they are fine but the data has dropped, this is where velocity earns its place. A measurable drop at a known load, on a reliable device, against an established profile, deserves a conversation.

Who should use it, and who shouldn’t

Use it if you work with athletes whose sport demands speed and availability, you have a device you trust, and you have time to build and maintain individual profiles. The review’s advice is to start where your environment allows: one-to-one and small-group coaches can run multi-point profiles, while coaches with large squads can use two-point profiles for upper-body lifts or published normative data.

Don’t bother if you coach large groups without measurement, you are in a high-volume accumulation phase, or your athletes aren’t technically competent enough for bar speed to mean anything. A novice’s velocity mostly measures coordination.

The general rule: VBT supports good coaching and doesn’t replace it. The device makes a coach’s judgement objective and shareable, but the judgement came first.

EPI delivers the National Certificate in Strength and Conditioning in Dublin and Belfast. For applying VBT across a competitive season, read programming velocity based training for team sport.

Frequently asked questions

What is velocity based training?

Using bar speed to prescribe and regulate load instead of fixed percentages of 1RM. Loading reflects how the athlete is performing on the day, and sets end at a defined velocity drop rather than a fixed rep count.

What is a velocity loss threshold?

A set percentage drop from the fastest rep in a set. The set ends when velocity falls below it. Research comparing 20% and 40% found similar strength gains with far fewer reps at 20%, and better preservation of jump performance and fast-twitch fibres.

Can velocity based training estimate my one-rep max?

Not reliably with a simple straight-line or two-point method on free weights. One study found that a quadratic model with at least four loads, predicting from the velocity at 80% 1RM, estimated the free-weight back squat 1RM within about 1 kg. Until that is replicated, direct 1RM testing is still the most accurate option.

Do I need an expensive device for VBT?

Device quality matters more than price. Linear position transducers generally show better validity and lower error than bar-mounted accelerometers and inertial sensors, several of which are only accurate at slow velocities.

Is velocity based training suitable for beginners?

Generally not. A novice’s velocity mostly reflects coordination rather than output, and the method assumes competent, repeatable technique.