There are dozens of ways to slice and dice the numbers with velocity based training.
We have so many to choose from: set average velocity and best rep velocity, mean velocity or peak velocity, plus there are secondary measures like tempo, various fatigue calculations, range of motion, and power... and that's all before we start talking about the profiling methods!
It can be pretty overwhelming, so in this article I will explore the key velocity metrics, and why I think the most popular of them are possibly the wrong ones.
The three most popular and widely used metrics are mean, peak and propulsive velocity metrics have their unique best use cases, but they also have significant limitations. I believe there is a superior and universal metric that no one is using yet, which can make all of three of these existing metrics redundant. Read on to learn how to measure it, and why I think this is such a big deal!
How the different velocity metrics are calculated
The velocity-time graph above shows the relationship between velocity at any given moment across the duration of a set. This example is shows five repetitions of a trapbar deadlift, and all reps have quite similar velocities.
The acceleration-time graph above is also from the same five reps of deadlifts. It's important to note that acceleration is describing a rate of change between time points, hence why it has such steep slopes as accelerations starts and stops on each repetition. This will become important shortly.
From these two graphs, you can see there are three different ways to calculate the the velocity of a repetition: peak velocity, mean velocity, and propulsive velocity.
The peak velocity for each rep is the fastest single moment of bar speed. Visually that is the highest point on the velocity-time graph for each repetition. In this set, each rep reaches a peak velocity of around 1.5m/s, with the last rep being closer to 1.3m/s.
Most velocity tracking technology uses a single time sample to calculate this peak, taking a snapshot from 1/100th of a second to determine this peak velocity.
Peak velocity is most commonly recommended for explosive exercises that have a float phase. Olympic lifts, throws and jumps differ from squats or bench press as the weight is accelerated maximally for the entire repetition ending in a release of the implement or when you leave the ground then experiences a period of float (or flight) when effort is no longer being applied. Peak velocity captures the single moment of greatest velocity and effort just before the release point.
In contrast, mean velocity measures the average velocity for the entire concentric portion of the lift. Beginning at the very start of the upward portion of the rep, through the fast portion in the middle, and also including the deceleration or float phase at the top of the rep.
Given a standard strength exercise can take anywhere from 0.5-1.5 seconds, mean velocity uses a number of data samples (50+) to provide a measurement. This increased sampling will be important shortly.
*You can also collect eccentric velocity data — which would be downward sloping portion on the graph for each rep — but that is a topic for another time.
Propulsive velocity, or mean propulsive velocity is the third most common metric used in velocity tracking. Propulsive velocity works much like mean velocity, however it differs in that it takes into account any active deceleration applied during the top of the rep.
For standard strength exercises with a load below 76% of 1RM, a lifter who tries to move with intent (as they should) must counter this explosive effort by actively decelerating the bar at the end of the range of motion to avoid releasing the implement or leaving the ground. Above ~76% of 1RM there is no active deceleration applied to the load, and propulsive velocity matches mean velocity exactly, below this threshold the lifter must apply a deceleration to the weight, the lighter the load, the greater the deceleration required (Sánchez-medina, 2010).
Basically, propulsive velocity is a better version of mean velocity, by actively filtering out the deceleration portion of a rep on light loads; it gives lifters a better picture of their lifting intent without being punished for applying the appropriate deceleration to the bar.
All three of these velocity variations (and any other velocity classification you can think of) can also be converted into a power metric by including bar load and the effect of gravity on that load. Power uses Watts (W) as the unit of measurement and can be expressed in absolute terms (W)or relative to bodyweight (W/kg).
The right measurement for the right situation
Some practitioners suggest alternating between peak velocity for your explosive movements and on light strength exercises (below 75% of 1RM) and then switching to mean velocity on the heavy sets (above 75%) however this switching creates issues with remembering to change settings, and becomes a problem for comparing data over time or accurately profiling with multiple sets.
Other schools of thought suggest that propulsive velocity be used for all strength exercises no matter what the load, but still suggest switching to peak velocity metrics for your explosive movements.
I think both approaches are flawed. The need for two metrics creates unnecessary confusion for new VBT practitioners, and makes comparisons between movements more difficult.
But there is a bigger issue at play, peak velocity itself is a seriously problematic data point due to it's data reliability and sensitivity issues.
Peak velocity: The unreliable and overly sensitive metric
Peak velocity — and by extension peak power — when compared with mean and mean propulsive metrics, is an incredibly noisy and overly sensitive data point. As a result it lacks both reliability and validity.
Whenever we measure something, we are looking for data to be both valid and reliable. These two factors are what make the data meaningful.
Validity is simply whether or not the numbers you are collecting actually measure what we think we are measuring. In VBT you can assess validity by comparing your numbers with a technology we know to be true, such as 3D motion capture. If a 3D motion capture setup says the barbell is moving at 0.68m/s and the VBT device outputs 0.68m/s, then we can say that data is "100% valid".
Reliability is the challenge of performing that measurement consistently from rep to rep, set to set, and session to session. If the data from one set is valid, but then the data for the next set is 20% out, we cannot trust that measurement. It's no better than just guessing the velocity.
Counter to what most people would think, reliability is more important than validity for implementing good velocity based training and programming.
It doesn't actually matter too much if your measurement is 10% below reality, as long as they are consistently 10% below reality. The data might always be "wrong" but if it is wrong by the same amount every day the trends we care about as coaches will be reflective of real changes to the athlete's abilities. We might not be able to compare our athletes to academic research or normative data, but this invalid but reliable tool is still going to be useful.
As an inverse example, if the data is 10% above reality on one day and 10% below reality the next this is now a useless tool. We cannot possibly use that device to provide meaningful insights into performance.
My problem with peak velocity is that it has poor reliability.
Because peak velocity data is taken from very few samples over a tiny portion of each repetition, it is very easily manipulated. A small pop, flick, or vibration of the bar, or a slight change in sensor placement can be all it takes to output an artificially high or low peak velocity measurement.
These small variations lead to unreliable data over time. Potentially hiding fatigue or poor lifting technique behind a wall of overly sensitive fluctuations and confounding variables.
Accelerometers in particular are prone to falsely high peak velocities due to the tiniest of bar vibration or altered sensor placement.
On the other hand, smoother averaged velocity measures like mean and propulsive velocity are more reliable as the greater number of samples for each velocity reading cancels out any noisy readings.
These metrics are very hard to manipulate or cheat, as long as technique and range of motion are consistent — things that can be readily assessed from video or with coaching. Mean metrics are strong objective indicators of a lifter's intent to move, or readiness/fatigue status over time. If mean velocity is significantly lower today compared to recent efforts (say -10%) on the same exercise and with the same technique, then readiness and intent are likely to be low.
The benefits of mean and propulsive metrics quickly becomes a limitation when it comes to measuring explosive exercises with a float phase. The longer the float phase, the slower the mean velocity over the entire effort. Mean velocity and mean power no longer rewards better reps with Olympic lifting or jumping. Propulsive velocity as described in the research is closer to solving the float phase issue however it requires an active deceleration to be applied, something that is all but impossible on these movements as gravity does all the decelerating for us.
We need a new way to calculate velocity in training.
An improved propulsive velocity - Working phase velocity
The original reason for creating a propulsive velocity metric was to ignore active decelerations applied during strength exercises with light loads.
This was a brilliant and significant improvement on mean velocity when dealing with lifting intent on strength exercises. Especially when recording velocity on warm-up sets, it has been a crucial application of VBT for collecting profiles and readiness data.
Unfortunately, because propulsive velocity requires an active deceleration to be applied, it fails to accurately capture the intent portion in explosive movement, which rely on the passive deceleration provided by gravity.
So I started to reimagine propulsive velocity, slightly tweaking the rules of how the propulsive portion is defined. I begun by simply naming it what I wanted to capture: working phase velocity.
This version of velocity aims to only capture the portion of a movement where effort is being applied to the implement in the concentric portion of a rep.
Instead of waiting for an active deceleration greater than gravity to be applied to the bar, working phase velocity ends when either:
- when the implement enters a float phase and is only being effected by gravity at an acceleration of -1g (-9.81m/s/s), or
- the movement ceases to move upwards (when velocity equals 0m/s), whichever comes first
This works for all types of movements, and will actually deliver surprisingly similar numbers to those you are likely already working with.
- For strength movements above 75% of 1RM this will be identical to mean velocity. These movements have negligible float phase or active deceleration so the working phase velocity will cover the entire concentric portion of the lift.
- For strength movements below 75% of 1RM it will be very similar to the existing version of propulsive velocity. These movements have an active deceleration so the velocity will only include the concentric portion until just as you hit the brakes.
- For explosive movements it will be somewhere between propulsive velocity and peak velocity. These movements have a float phase, so velocity data will be recorded for the portion of the movement where you are working to overcome gravity, this is from the very start of the movement until just before you leave the ground or release the bar.
There is something simple and satisfying about a universal metric that can be used for all velocity based training contexts.
There are a lot of benefits of having a universal metric for example it drastically reduces the complexity of training with and coaching with velocity and makes it much easier to compare performance over time or between exercises and across loads. It seems small, but being able to simply call it "velocity" and not having the hassle of changing settings in an app or transcribing the wrong numbers into spreadsheets reduces data errors drastically and makes for one less thing to have to educate athletes and interns on when they start with VBT.
Using working phase velocity may not capture the single moment of explosive effort for power movements, but what you might lose in validity you gain back in reliability and clarity. My hope is a single metric like working phase velocity can make VBT easier for people to understand and use.
But what do you think?
Let me know your thoughts, or experiences, do you agree or disagree?