well done article.

I'd like to have seen a comparative chart for cycling on the effects of cadence on power and Vo2, right?

in the end, a cycling power meter is about pointless unless correlated with Vo2 and mm/l of lactate, so if 1 "watt" with Stryd equals ~1 Kcal, like with cycling, does it really matter what the measurement proxy is for that calculation (deflection of metal vs an algorithm)?

---

wovebike.com | Wove on instagram

Interesting. Further cements my already pretty cementy disinclination to get a running "power" meter.

If the measurement is accurate, probably not, but at the same time I cannot imagine the slowtwitch crowd is going to be jumping on cycling power estimators like PowerPod.
--

The article is interesting for sure. I am intrigued by a running power meter, but not ready to adopt. One thing I wonder, and perhaps someone here can help explain, is whether or not increasing run power is always a good thing.

On a bike, I feel like once you control for conditions (wind, hills, road surface, etc) and aerodynamics (position, equipment, etc), increasing your power will result in an increase in speed, which is objectively a good thing. What about on the run? Does increasing your power equate to increasing your speed? For example, I can run at 8 min per mile without exerting a ton of effort (based on perceived exertion). However if I were to run at 8 min per mile trying to increase my vertical oscillation as much as possible (essentially jumping as high as possible on each step) I would not be any faster, but I would be exerting more effort. Would my power increase in this case?

Form plays much more of a role in running than cycling from my understanding. What I wonder about is are there ways to change your running form that will increase your power that will have a negative impact on performance (such as exaggerating your vertical oscillation in my example above)?

If so, could training to increase your run power potentially result in making form changes that make you a worse runner?

Perhaps I am way off the mark with this, but curious what others think.

Yes, it’s good thing in the long term, just like with cycling

Your mechanics example is equivalent to riding with an octagon for a bike wheel. Yes, rounding the wheel is smarter than putting out more power to compensate. And analogous to biking aerodynamics: get more aero before putting out more power

---

wovebike.com | Wove on instagram

Disclaimer: I have the original stryd, and my girlfriend has the latest one. I don't use them in day-to-day training (for a few reasons)

As a low cost indicator of fitness, I'll often run on a treadmill at 1% incline aiming for 140bpm and see what speed I get to. It's not perfect but I find it to be a good indicator of aerobic fitness. If I see an improvement, how do I know if it's because my cardio system is better or because I've lost a few pounds. I am *guessing* the power helps me triangulate that, but I haven't done enough testing to figure it out.

I found it more useful in the beginning when running hills. It was useful being able to smooth out the effort by aiming for constant power. It is much more responsive than my heart rate so it helped even out the effort across up hills and downhills. Once I had figured that out I didn't use it as much. I can't remember how it behaved in windy conditions. I'm light and if there are serious winds I do notice a different in pace going into or away from the wind. Not sure if running power can help you figure out if you're slow because it was windy. That's rarely a problem for running though, versus cycling. I also can't remember how it worked on sand. I did some isolated 'testing' and got bored before I found a breakthrough to be honest.

I don't think the power is useful for making form improvements. Form is too complex to bring down into a single metric. That's why we have ground contact time, steps per minute, etc. Even with all those I don't know you can just say to someone 'lessen your ground contact time and increase your steps per minute' and know whether their form has improved...

Most of my hard run sessions I just do on the treadmill which is a controlled environment so don't find power adds much insight.

The biggest difference in Cycling and running, is the degrees of freedom around which the power is exerted... In cycling you have a fixed axis that you are rotating the pedals around, so it's a more constrained movement (easier to measure, and more direct link between changes in force exertion and change in performance). In Running, you essentially have infinite ways (not quite, since ultimately it is limited by your flexibility and limb length) to deviate your form, which may or may not correlate with improved performance. In the learning literature (motor learning that is...) they refer to two concepts, knowledge of performance (kp) and knowledge of results (kr). In the context of running, knowledge of performance would be how you get to the finish line, and knowledge of results would be your finish time... When you look at mastery in this literature, essentially you see increases and plateaus in (Kr), but the argument that over the course of the plateaus (and inherently to overcome them) you have a change in kp (aka improvements in your performance that may not immediately improve the outcome, but are work towards this...). This is obviously simplified a bit in terms of explanation, but that's pretty much what it boils down to.

What does this mean in terms of running power meters...? Not sure. As the article suggests running is inherently harder to measure, likely impossible to directly measure. However, if the estimates are useful as a training tool, they could still be of value. I have yet to play around with one of the running PMs yet, so I haven't formulated an opinion. I did end up ordering one after a lengthy conversation with some exercise physiology experts on the weekend to play around with and see what we think...

In terms of the comment about the vertical oscillation and effect on power vs. speed (or economy for that matter) is probably not going to positively impact anything. While you could optimize vertical oscillation, increasing it at some point negatively affects horizontal displacement (while still requiring positive power to create the movement, you are effectively wasting the energy, since at a certain point it ceases to provide any benefit towards moving forwards), which is ultimate the goal of running...

Each running stride is effectively a series of accelerative movements (power movements, effectively toe off and knee drive) and recovery movements (foot contact and recovery (of the knee)). Effectively you generate acceleration with each stride, and if you find the optimal cadence, you effectively make the next accelerative movement, before you begin to decelerate (which allows you to maintain, or potentially increase the speed at which you are running). If you lengthen a stride for it's own sake, at some point in the stride you begin to decelerate, and have to exert more force to regain the same level of acceleration... Where the optimal balance point is will vary depending on your body (limb length, flexibility, bodymass, etc.) and external factors (wind, surface, slope, etc.). Can something like a running PM measure all of this? No. But can it provide a reasonable proxy of some of this, that I am not sure.

That being said, from the people that i have talked to who are using the running PMs, it tends to correlate well to VO2 as a measure of energy expenditure, and given the cost and invasiveness of portable VO2 devices (I've run with the pack devices, and driven with them on) it could be a reasonable option. When you factor in HR variability and cardiac drift, whatever composite the device ends up measuring, it is likely a less bad reflection of training intensity and estimation of the zones in which you are training, than running with an HRM... All this to say, I'll soon play around with the device to formulate a better opinion on it, but that it's unlikely to be as revolutionary as power meters have been to cycling, but it could still prove to be a valuable training tool.

External (gross or net) power is easily measured using a force plate. The only component not included in such measurements is wind resistance (which generally accounts for <5% of total).

Internal power, power recycling, etc., are all red herrings from a pure physics perspective. The only places runners interact with the environment are at the sole of their shoes and with the air molecules they push out of the way.

Where things get tricky are 1) when using accelerometers instead of a force plate, and 2) when interpreting the data (i.e., does running power even matter?).

I've had a Stryd for a year or so & I simply don't use it in that way.

For me there's a few useful things the Stryd does.
- Allows me to compare times on different courses (ie. hilly vs downhill/flat half-marathon), was I running at roughly the same effort? was it just the course that was faster? (ie. normalized graded pace on steroids)
- Power duration curve, seeing how that evolves over time
- TSS calculation based on more than just pace
- Instant power feedback in a race (so I can qualify my perception of effort)

It may not be as accurate as a cycling power meter, but it is useful in similar ways.

Genuinely curious, but everything you mentioned is taken care of by NGP. Can you elaborate a bit more on why running with power is "on steroids"? I've never used a running PM, so this is all very new to me!

Is NGP a live metric? Or do you only get it after uploading?

As far as I'm aware it only gets viewed after upload. That said, many head unit manufacturers do normalized power on the fly, so I don't see a reason why this couldn't be added to running watches (it may already be, but I'm on an old Timex Global Trainer).

They would need TrainingPeak's permission, since unlike normalized power the NGP algorithm is proprietary.

Another question might be which is more accurate, i.e., NGP or estimated running power? The former is designed to respond NON-linearly with respect to running speed/metabolic demand, whereas the latter is intended to track linearly with it.

I'm not certain of this but I believe NGP is calculated using GPS (or possibly barometric pressure) altitude data.

Plus as far as i'm aware there's no dynamic NGP field on Garmin's.

Interesting, I didn't know that NGP was proprietary to TrainingPeaks. I assumed it was just like Normalized Power where anyone could calculate it.

I’m bad at math... is it possible to back into an approximation mathematically of what would be applied to a forceplate using just a fancy accelerometer and my weight as Stryd claims, or are there too many variables ?

Sure, it's possible, and actually a pretty reasonable approach, as long as you understand the limitations of going from a recorded acceleration to an estimated "power" value. I say "power" since there seems to be some contention on whether the numerical value Stryd gives is representing power or not.

Here's Stryd's whitepaper which gives some insight into how Stryd compares to a force plate. Stryd's approximation is really close (95% claimed in whitepaper), however you can see where the force place is actually measuring the impact forces and the force oscillations (the large spike after foot contact and the subsequent vibrations) are observable as the running surface and foot interact. Stryd can't capture those force oscillations, or any forces at all, only the gross accelerations and rotations of your foot, so that's why the power is an estimate. In my opinion, a pretty good one and from my experience for the past 8 months, Stryd has been a useful tool.

An approximation, yes. The basic math is pretty simple.

I experimented with this in my garage a 5-6 years ago, modifying an aftermarket insole. Given I was just operating at a hobbyist level, the data from low-cost accelerometers is inherently pretty noisy. And I had huge issues with things like consistency/calibration/fixing the accelerometers well in a shoe.

I decided, at the time, it wasn't the best way. And switched, instead to a direct pressure sensor embedded in the sole instead. It would effectively measure the compressive force between ground and foot. I ended with a design that measured pressure at 5 points of the foot. Similar to what I think the Brim Bros. tried to do for a shoe-based cycling power meter.

I had much more luck with this approach. I decided the insole wasn't great (they get wet, etc.), so I converted into a "data sock" with a little battery/processor pod over the achilles.

Handed the package off to a guy who looked for venture capital type funding (this was before GoFundMe, etc), but he didn't have any luck.

I'm not claiming this is the best way. Maybe Stryd or others have figured out accelerometers, etc, and they're brilliant.

I also did very little in terms of processing the data. I didn't pretend that I'd created a "power meter." I showed the marketing guy raw data that showed we could very accurately calculate stride rate, and get some decent-looking data about integrating force over time on each foot strike to calculate the energy transfer. But I told him not to lie that it was anywhere near finished, and warned that it would take work on the order of several Ph.D. theses to arrive at something like a true power meter.

Apparently Dr. McGregor is smarter than I am.

I don't think the form thing is a big issue. It may vary slightly over time if you improve your form but it's not going to vary enough from one day to the next to impact your training or racing pace (unless you're consciously trying to run with bad form).

Bigger question for me is how much power is needed in running. The main need for measuring power as I see it is for scenarios where speed isn't an accurate measure of effort. Nobody worries about power in swimming, because you're swimming in fixed conditions, therefore lap speed is an accurate measure of effort (again form should contribute to speed increases over time but for a capable swimmer these changes will be gradual enough that using a % of recent race/test time is still accurate). Track cyclists use power, but not as much as their road counterparts, because track conditions don't change that much.

Road cyclists obviously use power a lot, and that's because wind, temperature, elevation, road surface, body position, group vs solo riding, etc all have a significant impact on speed, and they're too complex for the human brain to guesstimate them accurately.

Running is in the middle somewhere. Personally the vast majority of both my training and racing is on fairly flat, fairly good running surfaces. So paceis a pretty decent proxy for effort. On courses with some elevated/offroad sections, then I can use HR to pace those sections. HR may vary from day at a given pace on a flat surface, but if I'm targeting 8 minute miles on the flat, I've churned out a couple of miles with my HR in the low 160s, and then I hit a hill then I know that keeping my HR in the same ballpark is going to be similar effort on that day (next day it could be mid 150s or high 160s for the same effort). I guess if running power could be measured accurately then it would be useful to somebody who ran on some very variable courses. If running power is just an estimate, and/or you run on mostly fairly predictable running routes, just doesn't seem a need for it.

For sure force plates would be the means for direct measurement of external power (my comment was referring to doing that measurement in a portable way). But assembling arrays of force plates, or building the tech into a shoe would be extremely expensive (not to mention, that it would likely be very difficult if even possible to transfer from one shoe to another with the way the strain gauges would need to be built into the outsole) if even feasible (when we did shoe durability and absorption testing in the lab, we mounted the shoes to a prosthetic foot on a hydraulic sled, that would then repeatedly impact into a force plate to measure changes in force/absorption profiles over time...)...

It'll be interesting to see how this area evolves over the next few years.

No really. There are very cheap, durable and accurate pressure sensors out here. I made my prototype above for <$1000, which means manufactured volume price would likely be something way, way less than that. The trickiest part is finding pressure sensors with the right "dynamic range" and tuning them correctly.



Yes, but the same is true of accelerometers or just about any other sensor as things like the spring properties of the shoe and how tightly it fits on a given person affect just about any sensor. Some sort of calibration for a given person would be needed....

This sounds like the Sensoria sock (http://store.sensoriafitness.com/...-fitness-smart-socks). No affiliation, I don't own them or know anyone who does. Just FYI.

Maybe I'm an edge case, but I was a heel striker and I eventually suffered a series of uncommon injuries. Got myself some Newtons and worked on form, attempting to remodel to a fore foot striker. Seems to have worked. How do I know? Well, injuries are gone, but I also have data from the Milestone Pod (disclaimer - I own one and am in the process of buying two more and working with Milestone in conjunction with a class I'm teaching) that agrees - my 'old' form is mostly heel strike and my 'new' form (no longer need the Newtons - but I do use them for racing and training, intermittently) is mostly fore foot strike. Interesting that as I fatigue (say, more than 5-6 miles), I start to return to 'old habits' and my heel striking becomes more frequent.

So, not power-related, but new, accelerometer-based devices *can* provide useful form information. Unless your existing form is perfect, I guess! :)

Yeah, the marketing guy who was pitching our prototype was wondering if one of the venture capital people we pitched to snagged the idea, because this group popped up not too long after.

But I think it's probably unrelated...wearable sensors are not new, and socks are a natural fit. Probably just people with the same idea. (but who actually got funding of some sort :) )