Powerlifters Aren't Powerful, They're Forceful: a note on what we're measuring with velocity based training (spoiler: it's velocity)

• This whole article is esoteric. This is probably as far in depth I'm going to get with power because it really isn't a huge factor for powerlifting
• Even though there are strong correlations between barbell velocity and several other variables (load, intensity, exertion, your favorite skittles flavor, etc), you're not a Doctorate of all the Strong Things for using VBT. You're just another dude that collects numbers
• There is no gold standard of velocity based training's metrics. It's still kind of up for debate, especially if you're using it for things like estimating power, force, work, and capacitive magnetic reactance. Please don't use your VBT device for capacitive magnetic reactance
• Basically, the information we get from VBT, while potentially powerful, really isn't that large of a step forward than quantifying our understanding there's an inverse relationship between %1RM and the volume (sets X reps X weight). It's better than the traditional programming model (opinion), but not that much
• There are known knowns. Things that we know that we know. There are known unknowns. Things that we know that we don't know. But there are also unknown unknowns. Things that we don't know that we don't know. And ultimately: the absence of evidence is not the evidence of absence prolly means we're guessing
• I really don't follow up on the things that I plan to do, but here's a page worth of shit posting based on the things I've been reading. No more planned posts.

Introduction

In the first post of this series, I did several comparisons of 3 different sensors for variations of the squat, bench press, and deadlift. In that post, my comparison measure was 2D video analysis. The obvious red flag here is that of all the comparison measures - 2D video analysis sucks. But this brings into question as to what's an appropriate criterion measure, or gold standard of comparison, when we talk about velocity based training devices. Also, what about the other metrics most units pull besides velocity? How about force, power, and work?

Having these criterion measures is important. For sake of explanation, hydrostatic weighing is the gold standard for body composition (technically it's potassium counting, but we fucked that up with the nuclear age). Bioelectrical impedance, waist-neck approximations (the "Navy equation"), DXA, and BodPod are all compared to underwater weighing typically when evaluating them for accuracy and reliability. You can compare one of these methods to another besides hydrostatic weighing (IE: the navy equation vs bod pod), but the comparison always relates back to the criterion measure - underwater weighing. A good criterion measure will be highly accurate and reliable. Accurate meaning it measures close to the actual value and reliable meaning it will reproduce that value with relatively low variance.

Think of it like three weather men. One is always 10 degrees warmer than he forecasts, but it's consistent. The other one is usually within 5 degrees of the actual temperature, but could be lower or higher. The last one is always 5 degrees above. While he's as accuracte as the 2nd weatherman, he's also more consistant. Actually this is a horrible analogy. No one gets their weather forecast from the nightly news anymore. Moving along

It's a common remark that all you need to know about lifting you can learn in a physics textbook. Yet the same people that say these sorts of things have "power" days in their routine that aren't necessarily powerful or are basically flying in the face of some basic mechanics. Volume is a shorthand metric we use, as is intensity. The underlying variable for volume is work, and intensity is force relative to a maximum. These things aren't necessarily intuitive though, so we shorthand volume with the number of reps by weight, distance, laps, whatever.  Likewise, peak force isn't easily quantifiable, so we just use a ratio relative to our 1 repetition maximum. There's really nothing wrong with that until you start implying that variable resistance (bands, I'm talking about bands) accelerated as fast as possible has a meaningful difference from doing a 1RM. You really wouldn't know that any more than you would you know your altitude just by feel. It might sounds like I'm taking shots at the usual suspects, but really every coach needs to make some assumptions about mechanics as they relate to training because measuring all these things aren't meaningful. And ultimately, the body's response to training isn't exactly required to make sense. Sometimes it just works, albeit sometimes in spite of and not because of the training method. We need to be honest about our instruments' and methods' limitations. I do this all the time with VBT when I talk about deadlifts. Mean concentric velocity has less meaning with deadlifts.

So this will highlight the insufficiency of certain velocity based methods.

Video Analysis as a Criterion Measure of Velocity and Power

Video analysis is probably one of the old school reliables for velocity. Not the best, but has been the mainstay since before VHS. You can record something with a ruler in the frame and use the frame rate and distance of that scale to get a general idea of the rough edges of the movement. Folks have been doing this for sprinters, weightlifters, and all sorts of olympic sports for years. Nowadays, we've opened that up to the masses through things like Dartfish, Coach's Eye, Kinovea, and Tracker. These are all entry-level tools and very valuable.

What you lose in 2D video analysis is essentially depth. You can get horizontal and vertical distance, but getting a measurement outside of your initial measurement is sketchy. An easy way of thinking about this is weight plates. An Olympic weight plate is supposed to be 0.45 meters in diameter. So when analyzing bar speed, it's a pretty easy measure to use. Here's the information lost with 2D video though:

Ian Wilson courtesy of the Hook Grip instagram account.
Both of these plates are the same size when viewed in the same depth,
but it obviously fails when viewed in this objective

The key takeaways are to analyze objects in the same plane, film horizontally and vertically perpendicular to the plane you plan to measure, and have a known unit of measure (a ruler within frame, standardized disc sizes, etc). Some other issues come up concerning video quality and frames per second. Thirty frames per second are going to lose a lot of information from frame to frame, whereas 120 is going to capture a large amount. Likewise, better video quality is going to give you clearer edges to measure. Some phones that film in slow motion are capable of capturing 240 frames per second at a lower quality, so you have to manage clarity of the video against the number of frames you need. The right way to go about this would be to buy something that puts GoPro cameras to shame, but that's not a realistic solution for coaches living in the real world.

Now it's important to note that programs like Kinovea have been validated when it comes to simple movements, and we can describe powerlifting as simple movements. We can postulate about how the bench press is a full body movement and how all three lifts hit all of the things and yada yada. It's simple. There's an upward motion and a downward motion. That's simple. We only make it complicated by focusing on the minutia. Powerlifting is stupid simple, stupid.

A step up would be 3D motion analysis. Using an array of cameras about a stage, we can track movement in three dimensions. This helps understand more complicated movements as well as sort out the axes that affect rather simple movements. Much like with 2D video analysis, you're going to have a known unit of measure within the frame during initial calibration. This helps the software understand position, length, and movement reliably. The great advantage of 3D video analysis is you can track multiple points of interest. If you're solely interested in barbell velocity, you can place a reflective marker at different points along the barbell and body to track that. So not only can you track velocity of the barbell, but you can see the angular acceleration of each joint during phases of the movement. That angular acceleration of all the joints in the movement translates to vertical and horizontal movement by the bar.

Video analysis of this scale isn't practitioner friendly. It's cost prohibitive, technically demanding, and largely just more trouble than it's worth unless you're doing research. On the research front, Lake et al have heralded the use of a force plate and video analysis as a criterion measure to determine power. The force plate measures force and the video analysis measure velocity, so you get the all the components you need. Sort of.

Let's bring it back to barbells though, since that's the sort of thing powerlifting does. The issue with barbell based VBT is that bar speed isn't necessarily the velocity of the center of mass. There's debate on how you determine that, but this shows that divergence:

Lake et al's article gets straight to the business with the title: Barbell kinematics should not be used to estimate power output applied to the Barbell-and-body system center of mass during lower-body resistance exercise. I'll refer back to this later.

But how consequential is that? Volume isn't a true measure of work and intensity isn't a true measure of force, but they have been good enough to be usable. Especially when it comes to powerlifting, the difference between system center of mass (the barbell and body system) becomes less consequential the heavier you lift. As a male, your center of mass is near your belly button, and the more plates you stack on, the closer it will move towards the position of the bar. And that center of mass is going to change not just with load, but the position of your body. When this is consequential is when we're dealing with light loads: such as those done to develop power. It's a good thing the objective of powerlifting isn't about demonstrating the most power. You're really getting a metric of barbell power with most units and rough approximations of system (bar and body) for others.

One of my favorite past-times is meme-ing fitness industry stills from YouTube

Bar Tethers

Linear position transducers (what I keep calling tethered VBT devices), are the most common measurement of bar speed. Most device validations are in comparison to other, older LPTs. Some of the most common comparisons seem to be to Optitrak, Tendo, and T-Force. Just because it's the most common doesn't necessarily mean it's the most valid though. We're essentially comparing one device's yardstick to another's and seeing how close they are to one another and the frequency that the difference is the same. As long as we can keep comparing to that same yardstick, it doesn't matter. Ideally, you want accuracy and reliability, but short of that reliability across all your feasible working intensities is most important. So if a device sucks at measuring slow things, heavy things move slow, and you lift heavy things - don't do it.

Some proponents of VBT have heralded tethering devices as the gold standard. Anyone that says that is probably trying to sell a product or obfuscate what their device is measuring. As I demonstrated in the first article, it seems to be the case that tethered devices are more accurate and reliable for measuring velocity. It does not measure power or force though. To measure those things, you need force plates. You can calculate it with known masses (bar weight, body weight, etc), but it's essentially best for velocity. Everything else is approximation and calculation, and there is certainly some drift across different ranges.

I don't know how to say this better than the original author, so here it is:

"Other studies have indicated that methods that depend only on kinematic and kinetic results have limitations when used to determine power output (Cormie et al., 2007b; Hori et al. 2007). It seems that the linear position transducer technique overestimates power due to increased force output production derived from double differentiation of bar displacement. When this technique is applied also to the mass of the subject, standard biomechanical procedures are rejected in that force is determined without considering the acceleration produced through a movement (Dugan et al., 2004). Despite this limitation, our results indicate that monitoring bar velocity is a useful procedure to control load intensity in resistance exercises, as observed by other authors (Hori et al., 2007; McBride et al., 2011)" - Garnacho-Castaño

The Cormie et al paper from 2007 cited has an interesting experimental method shown here:

The key part here is the use of a force plate. The tethers are inverse mounted and it just does trigonometry to determine position across time. Both this method and the method demonstrated by Lake involved force plates. There's really no getting around actual measures of force. To put it another way, trying to use VBT devices to measure force or power is like using a thermometer to measure rainfall. Ok, not that bad, but sort of.

Another point worth mentioning is that the trigonometry used with two tethers is achieved all the same using GymAware's technology. Rather than measuring the lengths of the side of the triangle (with one unchanging known side length) determining the angles within, GymAware measures one side and an angle which gives you a measure of vertical and horizontal position (the other two sides of the triangle). All that shade you been throwing on social about trig, and now it's actually kinda useful ain't it? Trig is useful folks. Build a staircase.

It's important to point out that Cormie et al and Lake et al's method do stand in conflict to one another. To some extent, Lake is probably right that center of mass is the "real" variable we should be measuring (kind of like how work is the "real" variable for volume). At the same time, Lake does not sufficiently explain how they determine the center of mass. This isn't necessarily impossible either if you apply something like segmental densitometry (for example, DEXA gives limb weights). Without taking sides here (Lake vs Cormie), the important parts are 1) both of them point out the necessity of a direct measure with a force plate 2) both research groups have authors that have done a lot of research on powerlifters, so they're pretty damn rad.

Accelerometers

Accelerometers seem to bridge this gap to some extent, when the right inputs are used. The issue is Force. Force is a product of mass and acceleration. If we return to the issue as mentioned by Garnacho-Castaño et al, inertial sensors would seem a likely solution. If I haven't hammered the point home yet, while accelerometer technology has failed to perform on some fronts, it could be the missing link we're looking for. This is especially pronounced with power movements.

I've talked before about how you could theoretically determine velocity when an accelerometer is attached to a fixed point if you know the lifter's height. It's a rough approximation and relies on relatively normal proportions of limb length according to height and the mechanics of exercises being relatively comparable across a broad population. That method would require a certain amount of machine learning or constants and manual inputs of the lifter. At a minimum, I would think height and weight would be important to determine force and power. That said, if tethered devices clearly excel at measuring bar velocity, then they're probably the right tool for the right job.

Banyard et al down at Edith Cowan Uni used the same method that Cormie et al used and took GymAware and PUSH head to head in one of the most comprehensive comparisons. Again, these comparisons are made with a dual tether and force plate system:

Notice this is only figure 3 and 8. You should see the rest of them. Some have insisted that the difference between actual power and GymAware's power is only 10 watts. I'm seeing 100 watts. Follow the link.

If your main take away from this is that PUSH is hot garbage and GymAware is overpriced because it measures what it says it does, you're missing the point. The point here is that I'm not sure why these are currently two separate products. A combined approach using both a tether and an accelerometer could still be affordable (assuming GymAware doesn't set the price), you'd likely see additive value. Force plates are almost out of the question.  In order to cover the size you'd need for a simple movement like a squat, you're talking about an upfront investment of $600 minimum.

Just do the math. If the cheapest tether device is $200, and a force plate set up would cost $600, that's like 4 red competition plates you can't buy.

I haven't been paying terribly close attention to PUSH, but it would appear they're actually serving their market pretty well. They've implemented a new system that uses some video processing in combination with the sensor to collect different metrics. Again, this really doesn't matter for our purposes because, well, barbells. If you are using the PUSH as a powerlifter, one great feature they've implemented is velocity cut off notification. That makes it decent for hypertrophy work (because lighter=faster), but maybe not a strength. The only other VBT device I can think of that has attempted to add another input (IE: video, displacement, etc) is the Form Sensor. However, Form's altimeter really only helps to improve vertical velocity as far as I know.

Approximations aren't all that bad

Some accelerometer systems are getting their estimates of power based on the combined input of load and body weight. This sort of works, but we have to recognize it is an approximation. The issue with most accelerometers in the end is that they tend to be thrown off by sudden changes in direction, jarring not related to the movement, and Germanic talismen inscribed with the correct order of runic scripture. Ignoring some of the inadequacies there, approximations aren't too bad. 

A good example of combined measurement and constant (manual input) measures is the Powerlift App. When you break it down, the app functions as a video stop watch. It takes a video that you use as a visual reference point, you input a range of motion for that movement (which hopefully you measure appropriately), and choose start+end points for the concentric part of a movement. It's actually pretty good at it and has been validated in a lab. All the app does is convert the start+end point frame count to determine time, dividing the ROM by that time to determine mean velocity. You're not going to get peak velocity.

Side note: Powerlift is unlike many devices in that it actually stores correlations of load and velocity, as well as many other useful funtions. Most device manufacturers could take some tips from Carlos. Technically some do this, but is a sort of way that can only be reference manually - far from optimal. 

Again, with combined measures and inputs, you can improve this further. For example, if you were to take the average of your ROM from all loads over 90%, you could improve reliability near 1RM distance. You could also manually input your MVT in advanced options, but for sake of fidelity it's probably better to only use ROM input from another device you trust and video an MVT rep to keep it relative. 

Why it Matters: Science, Theory, and Methods

When I say the science, I really mean more of the mechanics - the sort of rules that things obey. There is something troublesome about scientific training. Firstly there's the fact the science does sort of exist in a vacuum. The protocols employed by science are commonly used to elicit a response, to tease out a marked difference or similarity of one treatment versus another. It really doesn't matter if it's a realistic training program (let's pretend that's the only kind of sports science for this), but one that emulates enough elements of real training protocols but has a magnitude large enough to elicit a response. Basically, most of these programs don't have ecological validity - they aren't programs that would be optimal.

Training theory on the other hand (not "scientific theory") is basically the rationale which we think training follows. This is a blend of what coaches have seen in practice and lab rats have evaluated in a laboratory setting. That connection from A to B isn't always a straight line. For example, the force-velocity curve is more based on the functional characteristics of muscle fibers. It's not necessarily the same thing when we're talking about movement created by multiple muscles and about multiple joints.

Lastly, the methods are how we choose to execute our training theories, hopefully based on the practical applications of the science. Using bands is one such methodology. I've beaten that dead horse long enough though, so let's use Orange Theory Fitness. Orange Theory Fitness markets itself as a science-based weight loss program. It sort of is. If we think of work (joules) as a proxy for energy expenditure (calories), then high volumes of energy expenditure are more obtained at lower intensities. I'm basically saying pure cardio. That's not exactly the Orange Theory Fitness training program. Secondly, there's a significantly greater magnitude of difference obtained through diet. So it's not all bullshit, but science did not divine the right answer to Orange Theory fitness and there's plenty of caveats. But science sells. Even diet programs use that tactic.

Everyone fails at marrying these three elements together, including your gurus, to some degree. With velocity based training, most of that fault comes from the science. I spent this whole article highlighting disagreement. The training theory failing comes from over-expansion of concepts like the force-velocity curve (IE: the training zones). An extension of that carries over into methods. Assuming the force-velocity curve can be made so specific in gross movements could be made even worse with the idea when adding variable resistance.

Putting it All Together

If you own a tether system, stop being so uppity. You have the most appropriate system for measuring velocity, and even that is still subject to error - even more so if it doesn't measure angle of pull. You could probably rig up two single axis system to improve vertical velocity measurement and maybe even track bar position throughout a movement, but there is no consumer level solution yet. Beyond the fact that tethers aren't appropriate for movements in three planes of motion, accelerometers do have additive value to help us better measure simple training metrics - like force, power, and work. We don't have those solutions yet, but that's likely the future. It might seem ridiculous to forecast the future like that, but given the relative utility of Accelerometer-GPS units for field practice, it's not too far off.

Velocity based training is supposed to be a method to quantify things we were making guesses about. Generally, something that's powerful was subjectively fast enough and light enough. I say subjectively because it has to be someone with some authority and experience in developing powerful athletes with movements that best mediate that training quality. That sense of what is powerful is largely a presumption. As much as velocity tells us, there's still plenty we can't nail down. Operating in this unknown area is fine though. We've been operating in the unknown using simple measures for decades. But you should probably stop taking VBT measures of power and force at face value.

#MaybeJustLiftWeights #ButHaveYouReadSUPERTRAINING