Mean vs Peak Velocity in VBT: Which Metric Should You Use?

Not all velocity numbers mean the same thing — even when they come from the same lift on the same day. Across the velocity-based training (VBT) literature, researchers report three distinct metrics: mean velocity (MV), mean propulsive velocity (MPV), and peak velocity (PV). On a heavy back squat near 85% 1RM, MV might read 0.44 m/s, MPV 0.51 m/s, and PV 0.90 m/s — three very different numbers describing a single repetition. Understanding why those numbers differ, and which one to trust for load prescription, fatigue monitoring, and strength assessment, is fundamental to practising VBT correctly. Sánchez-Medina et al. (2010) demonstrated that MV and MPV follow highly linear load-velocity relationships in the squat and bench press (R² > 0.95), yet the two metrics behave differently as load increases — a distinction that carries real consequences for monitoring accuracy.

Each metric samples a different portion of the barbell's velocity-time curve during the concentric phase of a lift.

Mean Velocity (MV) — also called mean concentric velocity (MCV) — is the average velocity computed across the entire upward stroke, from the moment the bar begins moving concentrically to the moment it decelerates below a device-specific zero crossing or reversal threshold. Every sample point from lift-off to lockout is included in the average, including the deceleration zone that appears at the top of most compound lifts when muscular effort tapers before the bar stops.

Mean Propulsive Velocity (MPV) limits the average to only the propulsive phase — the portion of the concentric stroke during which the barbell is still accelerating, or more precisely, where net barbell acceleration remains positive (above −9.81 m/s²). The deceleration phase, where the body must actively brake the bar to prevent hyperextension, is excluded. The propulsive phase fraction is not fixed; it shrinks as absolute load increases.

Peak Velocity (PV) is simply the maximum instantaneous velocity observed at any single sample point during the concentric phase. In most lifts it occurs somewhere near the mid-concentric range, well before the deceleration zone. PV is sensitive to the power peak of the movement and captures the highest expression of neuromuscular output within that repetition.

The core utility of any velocity metric in VBT is its ability to accurately predict or track relative load (%1RM). A reliable load-velocity profile lets coaches prescribe training intensity by velocity target rather than a percentage of a tested maximum — a powerful advantage when daily readiness fluctuates.

Sánchez-Medina et al. (2010) published what remains one of the most cited load-velocity profiles for the full squat and bench press, establishing that MV and MPV both yield strong linear relationships with %1RM across a range of 40–100% 1RM. However, the two metrics differ in how stable their reference values are as load approaches maximal.

At low to moderate loads (40–70% 1RM) performed with maximal intent, MV and MPV diverge considerably because the deceleration phase occupies a larger fraction of the lift at submaximal loads — the bar overshoots the force requirement and must be braked. At these loads, MPV is systematically higher than MV. As load increases toward 1RM, the propulsive phase extends, the deceleration fraction shrinks, and MV and MPV converge. Sánchez-Medina & González-Badillo (2011) confirmed that at very heavy loads (above ~80% 1RM), MV and MPV become nearly equivalent.

PV is always the highest reported value for a given repetition. Its load-velocity profile is also reasonably linear, but the slope is steeper and the absolute values at 1RM are higher than for MV or MPV. This means that PV-based velocity zones must be established separately from MV/MPV zones and cannot be used interchangeably.

MPV was originally proposed as a solution to a specific measurement problem that arises at submaximal loads. When an athlete lifts a light-to-moderate barbell with maximal effort, the muscles generate substantially more force than required to move the load, causing the bar to accelerate strongly in the early concentric phase. By the time the bar reaches the top of the range of motion, the lifter must actively decelerate to avoid hyperextending joints or losing control. This braking effort contaminates the MV calculation: the slower velocities recorded during deceleration pull the average down, making the lift look slower — and therefore heavier — than it truly was in terms of effortful output.

Sánchez-Medina et al. (2010) demonstrated that MPV corrects for this contamination by excluding the deceleration phase. Their data showed that MPV produced tighter, more consistent load-velocity relationships at submaximal intensities compared to MV. The argument is compelling: if you want a velocity metric that reflects the athlete's actual force-production effort rather than a diluted average that includes obligatory braking, MPV is theoretically superior for light-to-moderate loads.

The practical limitation is that not all devices compute MPV in the same way. The threshold for identifying the end of the propulsive phase (typically where acceleration crosses −9.81 m/s²) must be calculated from second-derivative data, which requires either a high-resolution accelerometer or a linear encoder with sufficiently high sampling frequency. Positional transducers with lower sampling rates may underestimate or misidentify the propulsive fraction, introducing device-specific errors into the MPV calculation.

Reliability refers to the consistency of a measurement across repeated trials under identical conditions. Sensitivity refers to the metric's ability to detect small but real changes in performance — what statisticians call the signal-to-noise ratio or the minimum detectable change.

García-Ramos et al. (2018) compared the reliability of MV, MPV, and PV during the bench press across a spectrum of loads and found that all three metrics demonstrated acceptable intraclass correlation coefficients (ICC > 0.90) under standardized conditions. However, PV showed meaningfully higher within-session variability (coefficient of variation 4–7%) compared to MV and MPV (CV 2–4%) across most loads tested. This higher variability occurs because PV captures a single instantaneous peak that is sensitive to minor technical fluctuations — a slight timing difference in where peak force is expressed can shift PV by several centimetres per second without reflecting any true change in the athlete's fitness or readiness.

MV and MPV tend to show lower CV values, particularly at moderate-to-heavy loads, making them more sensitive metrics for detecting genuine day-to-day or session-to-session changes in neuromuscular output. When the goal is daily readiness monitoring — deciding whether an athlete is fresh or fatigued relative to their normal load-velocity curve — this stability matters considerably.

Pérez-Castilla et al. (2019) examined the sensitivity of each metric for detecting fatigue-induced velocity loss during high-volume squat protocols. Their findings reinforced that MV and MPV tracked fatigue progression more smoothly and with less noise than PV, supporting their use in velocity-loss thresholds (e.g., stopping a set when velocity drops 20% from the first-repetition value).

The following table summarises the key practical characteristics of each velocity metric across the dimensions most relevant to VBT practitioners.

At loads above approximately 80% 1RM, MV and MPV converge to nearly identical values, so the choice between them becomes less meaningful for heavy strength work. The most impactful distinction lies in the 40–75% 1RM range and in ballistic versus controlled-concentric exercises.

Despite its higher variability, PV is the appropriate metric for several important training contexts — and using MV or MPV in these situations would actually be less informative.

Ballistic and jump-based exercises are the clearest case. In a jump squat, hex bar jump, or countermovement jump, the athlete deliberately leaves the ground at the end of the concentric phase. There is no deceleration phase within the movement because the purpose of the exercise is to project the body or bar upward as fast as possible. The entire concentric stroke is propulsive. In this context, MV and PV are equivalent in their meaning, but PV is often reported because it captures the highest expression of elastic energy storage and neural drive during the most powerful moment of the movement.

For ballistic upper-body exercises such as the bench throw or medicine ball chest pass, PV is similarly the standard metric because athletes release the implement at the point of peak velocity, making PV the direct measure of mechanical output that translates to projectile performance.

Light-load power assessments (below ~40% 1RM) also benefit from PV reporting. At very light loads, the bar decelerates over a long portion of the lift, and MV is substantially contaminated. While MPV theoretically corrects for this, PV provides a simpler and often more reproducible representation of the peak power window. Researchers investigating force-velocity profiles that extend into the velocity-dominated end of the spectrum typically report PV for this reason.

Monitoring explosive ability independently of strength is a legitimate use case for PV. An athlete's PV during a loaded jump squat at a fixed submaximal weight can track their rate of force development and elastic qualities over a training cycle more directly than MV, because PV at light loads is more sensitive to the rate of force development than to maximal strength expression.

Given the evidence above, the following guidance helps practitioners select the right metric and use it consistently.

1. Match the metric to the exercise type. For controlled compound lifts (squat, bench press, deadlift, Romanian deadlift) across moderate-to-heavy loads, MV and MPV are both appropriate. At submaximal loads where you want to capture true effort quality, prefer MPV if your device computes it accurately. For ballistic movements (jump squats, hang cleans, hex bar jumps, throws), use PV.

2. Establish your load-velocity profiles on your device. Do not import velocity targets from a paper that used a different device. Build an individual load-velocity profile for each athlete on the specific equipment you use, recording either MV or MPV consistently. A three-to-five point profile across 50–95% 1RM takes approximately 20 minutes per athlete and generates individualized velocity targets that are far more accurate than population averages.

3. Pick one metric and stick with it. The single most common error in applied VBT is mixing metrics across sessions — comparing today's MV to last week's MPV or to a PV from a different device. The absolute values differ substantially; only consistent use of the same metric on the same device produces interpretable trend data. If you switch devices or decide to change metrics, re-establish your profiles from scratch.

4. Use velocity-loss thresholds with MV or MPV. When autoregulating set volume through velocity loss (e.g., terminating a set when velocity drops 20% from the first repetition), MV and MPV are both appropriate and preferable to PV, given their lower variability. PV's higher CV means that a 20% velocity-loss threshold using PV is less reliable as an index of neuromuscular fatigue.

5. Treat daily monitoring VBT-readiness assessments with MV at a fixed submaximal load. The standard readiness protocol — an unloaded or lightly loaded reference set at the start of each session — works best with MV because of its lower noise and straightforward interpretation. If the athlete's MV at the reference load has dropped more than one standard deviation below their moving average, it is a signal to reduce volume or intensity that session.

6. Don't ignore PV for athlete profiling. Even if MV or MPV is your primary training metric, tracking PV during ballistic exercises over a training block provides complementary information about explosive power development that MV-only monitoring misses. A complete VBT picture uses MV for strength and loading decisions, and PV for explosive power tracking.