I've never had an issue with gps for distance. I regularly run an out and back course, and although instant pace is patchy, If I turn back after 4 miles say, my watch will log 8 pretty much within a couple of feet from where I set of. I've had the same experience with pretty much any Garmin I've tried, starting with the 305, then 620, 910xt and now Fenix 3.

I know people complain about accuracy, but I've never had an issue where I live.

After 11 months with the pod I'd completely forgotten how poor non-Stryd instantaneous pace was.

Can you explain how this data shows the accuracy of the Stryd compared to VO2?
From what I see in the graph a VO2 of 45 can give a Watts estimate between 2.9 and 3.8 w/kg at 55 it goes from 3.5 to 5.1. That is pretty large variability.
Does Stryd take into consideration RE somehow to make it more accurate?

---

Tiago
---------------------
Sponsors: : Blueseventy :

You work for Stryd and offered to answer questions, so, a really basic one. How is it logically possible that Stryd could offer highly accurate measurements of distance covered? For the sake of an example, how could a Stryd unit, with its associated sensors and algorithm, be extremely accurate in measuring the distance of a marathon-length run?

Disclaimer: I am with Stryd


Hi niccolo,

Thanks for your post. This is one of our breakthroughs. Stryd has extremely fine grained spatial and temporal motion tracking. This allows us to get highly accurate speed and distance without calibration. We track the whole motion of every stride during your run. Please see http://fellrnr.com/wiki/GPS_Accuracy for details surrounding Stryd's accuracy.


We are compatible with the Fenix 3 Sapphire and are dedicated to supporting as many watches as possible. Let me know if you are interested in trying out running power.


The paper jumps into an explanation of this right after the 'The Stryd data are just as good as the VO2" graph.

The authors dissect the following factors: metabolic efficiency (paper here), energy cost of running (paper here), and running economy (paper here). If anyone is interested in reading more, Ron's and Han's complete works can be found here.

Best,
Angus

So basically you're saying that the sensors in the Stryd allow it to so precisely track spatial dynamics that it can differentiate between, say, a 99, a 100 cm, or a 101 cm stride length. So the Stryd is literally measuring every step a runner takes, so precisely that all those steps add up to the total run distance significantly more precisely than the best consumer GPS tracker available would report. Honestly, if it weren't for the fellrnr.com data you mentioned (with which I was at least somewhat familiar and which seems credible), I would tell you that you were smoking something. I'm still struggling to wrap my brain around this...

It's really not that hard to fathom, this is how all footpods have been working like this for years not just Stryd.

They don't simply measure cadence and multiply that by a fixed stride length for your entire run. They utilise the accelerometers in the footpod to measure foot swing force and effectively use this as a dynamic calibration factor for measuring distance that can (to a certain degree) compensate for changes in gait.

As someone else posted above, if it was simply a case of multiplying cadence by a fixed stride length then the Garmin models with wrist based cadence would do just as good a job at measuring pace and they really don't.

I routinely use a footpod and just GPS for about 50% each for my runs and whilst the distance is usually +/- <1%, the major benefit of the footpod is more accurate instantaneous pace. And this is a Garmin footpod, not Stryd.

It's really not that revolutionary, you should join us in the 21st century :)

Revolutionary or not, I had no idea you could so precisely infer distance covered just from accelerometer data. Now I'm trying to wrap my head around whether this is unique to running or whether you could make similarly precise inferences in other activities (something like hiking presumably yes, swimming I'm not so sure).

It is basically inertial navigation, used in everything from subs to airplanes to missiles to spacecraft. The biggest issue with inertial navigation is drift, the longer you are using it, the longer your uncertainty in your position. All those little errors in acceleration and vector add up over time, resulting on your velocity and position being in correct. Fancier systems have less drift, but running a marathon requires only takes ~4 hours compared to a submarine that is under water for days traveling hundreds of miles under the polar ice caps. Due to drift all those platforms zero their position using another system like GPS or star position to negate their drift. I assume stryder is also using gyroscopes in addition to accelerometers.

In other words it can be used in every sport.

He's speaking from a trail running perspective. Heavy tree canopy coverage plays havoc with GPS reception. Depending on the time of year (ie, are there leaves on the trees), you can see swings of 15-30 seconds per mile. The signal drops--and then doesn't reacquire--then straightlines your 'path' instead of tracking all of the mileage. Plus--with the GPS only having so much resolution (it's what, accurate to within 3-5 meters these days?) a twisty path gets even more straightlined. I train on trails, have a Fenix 3 + footpod and recalibrate it every so often.

I don't think that's right. His baseline testing is on open roads, and the Stryd crushed the best available consumer GPS watch. As an additional test he subsequently tested it on trails. I don't have time to go back and look up the specifics, but I remember thinking the same thing you did, and then correcting myself after I looked back at the site.

Gotcha--however, he still puts the footnote down that he's testing on curvy roads---and he's ignoring data during the winter months with no leaves on the trees.

I've been running with Stryd since May and can vouch for fellrnr's findings. I've observed the same accuracy and repeatability with the foot pod that most people are seeing. I also run at a college track a couple of times a month to verify that Stryd is measuring good distances and look at the raw data over 10+ laps. Not every single Stryd user has had this experience, though, and some users do require some sort of calibration factor before using it for training, but from reading their forum and Facebook page, most users are seeing great accuracy and precision.

You have to kind of understand the underlying tech of GPS to know that it's inherently VERY inaccurate, especially for cheap consumer devices, GPS watches included. A GPS is estimating your location based on signals from satellites over 12,000 miles away. You have to spend several thousand dollars on a survey-quality GPS for reliable 1-meter accuracy. The foot pod is measuring your foot's movement, on your foot, and likely sampling the 9-axes of motion (translation, rotation, magnetic orientation) several times a second. If you research Arduino and the crazy amount of DIY sensors you can buy, you'll see how accurate cheap sensors can be at home. Now imagine how good the data can be when you have high quality sensors and an engineering team hard at work refining the device.

My data from a recent training run:
Lap, Distance, % Error - 400m
Lap 1: 402.2, 0.55
Lap 2: 401.9, 0.48
Lap 3: 405.6*, 1.40 - passed a couple walking on the inside lane
Lap 4: 401.2, 0.30
Lap 5: 403.4, 0.85
Lap 6: 400.8, 0.20

Power estimates and cost aside, it's really a reliable device. I have more than two decades of being a tech hobbyist and now an engineer who uses tech at work, and I think Stryd is a very unique device with a fantastic support team behind it. Other manufacturers should strive to provide the same or better trueness and repeatable performance, because it should matter to everyone who doesn't run on a track to collect the best data they can. Why not, right?

I think the root problem with his protocol, is that he has known 1/4 mile markers on a test route. He counts each 1/4 individually as a data point. He does not say whether he has multiple runs with a device on that route, or if he just does one run per device. I suspect that GPS tends to correct and refine itself over the course of an entire run. So, let's say a device shows 0.23 at the first quarter and 0.50 at the 2nd quarter. In that case, the first data point would be 0.23 and the 2nd data point for 0.27. So, on paper it looks really bad, but is accurate for the total distance. By contrast, say a pod or another device shows 0.24 on the first mile and 0.48 on the 2nd mile. In that case, it would appear more accurate, yet be less accurate overall. I suspect that is why pods show more accurate to him than GPS devices. My GPS devices have a very high degree of trueness and precision for total runs, where iPhones and pods are rather crappy.

There may be some problems with his protocol, but he seems very rigorous with his testing. There are lots of us who can claim the same results with our Stryds, though, and some of us are just as rigorous with our testing. Your example showing 0.24 for the first mile and 0.48 for the second mile would be for an uncalibrated Stryd. Once calibrated, realistically, it would be something like 0.246 miles and 0.255 miles to equal 0.501 miles. See my example above.

But you are mistaken that a GPS device refines anything during your run. It reads signals and triangulates your location with some degree of precision, and any accuracy you see would be a byproduct of cancelling errors, not true precision. Can GPS accurate over longer distances? Sure. Can you have a really repeatable GPS? Of course. I use survey-quality GPS at work with 10-cm resolution, and it's pretty great, but my fenix3, cell phone, and eTrex can't come close to that level of precision. And to get 10-cm, my system needs two GPS units, one as a base over a known benchmark (and 10+ minutes for DOP to settle), and one as a rover to communicate with the base. I'd do a test run to compare the survey GPS with Stryd, but I haven't figured out how to run with the antennae yet.

Can I ask how you can claim to know your GPS device has a very high degree of trueness and precision? In what conditions? How do you know? What data do you have? How have you measured and verified your distances? I collect and use data for a living, so I am always curious of people's test methods for verifying that their devices are performing correctly.

The point that I was trying to make is that every individual GPS data point has a margin of error. However, over the course of a run, it will collect a crapton of readings. The net is likely a very accurate overall distance, though internal distances between data points could vary more. I believe that I am seeing this in the statistical analysis I have done with my devices on my routes (and confirmed by two friends who have different devices on the same route).
I can speak to precision, but not trueness. I have hundreds of runs on a routine course I have done for over a decade and using multiple GPS devices. The mean distance is 5.07 miles with a standard deviation of 0.02 miles. All of the GPS watches have had the same results. Maybe one of the older watches was a hundredth of a mile different SD, but I cannot remember without recalculating. By contrast, I have dozens of data points with an Apple Watch and iPhone over the same route. The mean was 5.06, if I remember correctly, but the SD was 0.1 miles. I only have dozens of Apple data points, because it was apparent pretty quickly that they suck, and I quit using them for runs.

Given the volume of devices with the exact same (or very close) mean, I trust the trueness of these in general. But there is no way that I it is worthwhile to go out and measure it to the inch.

In summary, GPS devices = awesome precision; AW & iPhone = mediocre precision. I have not done any analytics with pods, but I have no doubt that a calibrated pod would return good results, but only so much as your gate is unchanged (Stryd excluded).

Angus, any chance Stryd will soon figure out how to make the footpod work with the Garmin 5s?

---

just your average age grouper . no one special . no scientific knowledge . just having fun.

In running there is an added factor that you know that 80ish times per minute, the foot becomes motionless for a fraction of a second, and you also know when that occurs.


I'm amazed that other folks are amazed at this, the first running watch I had with speed and distance was (I think) the freestyle fs-1, it used a footpod, and used what would later become ant+. That was 2002, there were other watches that worked the same way, I am reasonably confident that the nike elite triax was one.

The garmin forerunner 301 came out in 2003 and was less accurate than the freestyle watch.

This is the basis for inertial navigation systems, used for decades in ships, submarines, airplanes, and spacecraft. See https://en.wikipedia.org/...al_navigation_system. It's a pretty well-understood technology and can be surprisingly precise if you have good sensors.

That's an interesting point. Tho I think of how I use my PM on the bike. I'm looking for a drop in power to correlate with an increase in speed, or hold a particular output for a prolonged duration, etc. I imagine it would be the same for running: power would simply indicate performance in a particular context.

l'm still rocking a FR910xt so I'm not using Garmin's running dynamics and I don't have a Stryd pod. So I have no idea what level of analysis they provide. But, at the very least, if you layered power into metrics such as pace, HR, incline, etc., wouldn't you be able to see the correlation between power and overall efficiency?

Full disclosure: I too know next to nothing about run power.

EDIT: Apologies. Should've continued reading thru the thread. Looks like folks far more capable than me have cleared this up. No more posts till caffeine has kicked in.

Hi way2sloow,

Stryd is technically compatible. However, we do not currently recommend the Fenix 5S. There are widespread reports of ANT+ dropouts when using Garmin heart rate monitors, Garmin power meters, Garmin run dynamic pods, and 3rd party sensors that make the experience unusable.

Garmin has essentially stated that there are issues on the Fenix 5 and Fenix 5S that will only be fixed with new hardware.



If anyone ever has any issues with Stryd, please email us at support@stryd.com and we will take care of you.

Best,
Angus