j
Stryd Accuracy vs Stroller
A reader on this site, @AndersB, has agreed to share some of his Stryd Accuracy testing results which I found interesting as he was comparing them to a wheel sensor on a (kiddie) stroller. He also performed some tests comparing to the GPS tracks from V800, 310XT and Apple Watch (gen 6).
About his findings, he said this
I did not expect Stryd to be this good, judging from what I read online.
and this
I got it for more stable pace, but was blown away by its accuracy.
and this
the Stryd (relative) elevation is also spot on. I’ve compared Stryd, Edge520+ and V800 for a few runs, and they have all been in perfect agreement.
and this
The Stryd…has been extremely accurate for me, with a spot-on pace.
Methodology
@AndersB used a Wahoo RPM sensor on a Thule Cross Stroller which he states to be ‘carefully calibrated’ and compared results with an uncalibrated Stryd Wind pod, sometimes also with other devices.
- The Wahoo RPM wheel speed sensor has been carefully calibrated to a specific tire pressure and verified on a 400m track.
- The wheel speed sensor is mounted on one of the rear wheels. The front wheel is prone to lifting when turning and lifting slightly when bouncing
- The wheel speed is recorded to a Garmin Edge 520+
- Apple Watch 6, Polar V800 and Garmin 310XT were also used from time to time on the non-holding hand OR mounted to the stroller handle with the GPS receiver pointing at the sky
- Stryd was attached to either Nike Zoom Winflo or Asics Gel-Contend
- Paces varied between 10:00/km and 4:00/km
Results
Let’s start off with the distance data. Here the variation to the calibrated wheel stroller is shown against each device on test. Clearly, Stryd and the V800 are the best.
The first chart is the most interesting to me. Whilst the V800 and Stryd arrive at the same distance, the V800 gets there by cancelling out of over-recording and under-recording distance.
Now, looking at speed over one run
and speed over another run
These relatively small variations in pace won’t matter to everyone. But will matter to some.
A Polar Study
A Polar-supported study published in 2020 found Polar V800 to be the most accurate on test.
GPS Watch Accuracy: peer reviewed University Study says V800 best
Take Out
This was the first test I’d seen comparing Stryd to a stroller and I would be interested to see the tests repeated if @AndersB also calibrated the Stryd on his local running track.
Some strollers have smaller wheels and these might be more prone to several kinds of inaccuracy but the model used seems to give results that are probably indicative of reality.
People often forget the phenomenon of under-measuring and over-measuring during a run. In other GPS-only tests, I have done over 10 miles with pretty much every GPS watch ever, almost all the watches record overall distance to be accurate to +/1% and that is over a route with occasionally demanding reception conditions. Yet many of the watches are often 10m off course for extended periods.
I guess, none of this really matters if you only ever want to log total mileage …if you want to run 10 miles or 10.1 miles then what is the material difference? A: None. But the accuracy of every individual metre that you run IS critical if you want a chance of getting an accurate instant pace. Thus, if you want to push yourself to the limit to getting that 20:00/5K time but your watch varies between 3:45/km and 4:15/km then you are going to find the tech useless for instant pace…although your lap pace will be OK once you are a minute or so into each kilometre. But then, do you speed up or slow down when the lap pace shows 4:05/km? (Answer: You don’t know, think about it)
Reader-Powered Content
This content is not sponsored. It’s mostly me behind the labour of love which is this site and I appreciate everyone who follows, subscribes or Buys Me A Coffee ❤️ Alternatively please buy the reviewed product from my partners. Thank you! FTC: Affiliate Disclosure: Links pay commission. As an Amazon Associate, I earn from qualifying purchases.
Do you know the distance the test done over?
I’m trying to approximate the % error.
Hi Will, Anders here.
Most runs were 5km.
The longest were 8.5km, where V800 ended up with a diff of 48m (0.5%), Wahoo Element Bolt v2 (I know, its a cycling computer, I was just curios) was 131m off (1.5%) and was Stryd 3m off. On that run, the Stryd was however 8m off at about 5km and then drifted back.
I did analyse about 10 runs, and the one where Stryd had the biggest error was 20m over a 4km run (0.5%).
Regardless of the pretty accurate distance I had, what surprised me the most was just how spot-on the instant pace was.
I have no doubt the Stryd is more stable than an older watch on a good, flat road — especially in any unfavorable urban conditions for GPS. The problem here is a 310XT is 12 or 13 years old — maybe 5 or 6 technology generations out. The V800 is also an obsolete even if it was the reference standard at one time. The current multi-constellation + multiband GNSS available from Apple, Coros, and Garmin are a substantial improvement over any previous generation. My personal experience is with Garmin fenix 7X specifically and overall I would have a hard time recommending someone with an 310XT invest newly into the Stryd over just getting on with one of these new watches.
I have Stryd for years and I still rely more on heart rate zones than power zones for most of my training. (I think power zones shine for targeting effort on intervals, especially under 1km.)
I have a few Stryd annoyances/concerns/caveats:
1. It seems to struggle with pace/distance with hills
2. Uncalibrated distance reads about 1.5-5% short on both Stryd pods I have had (around 1.5-2% on the original and 3-5% on the Stryd Wind) but is also substantially different between shoes.
3. I think the power readings are a bit wonky on hills both up and down compared to my RPE but I cannot really measure it.
4. It seems sensitive to changes in terrain from tarmac to broken ground to grass to dirt to gravel.
I think these may be issues inherent in the technology. I find the watch in multi-constellation, dual-band GNSS is more consistent across shoes and terrain for distance and so I have the watch auto-calibrating the Stryd which you aren’t really supposed to do, but equally I am not going to calibrate each pair of trainers. I also think that I trust the Garmin SatIQ for distance more than Styd in general when the terrain is mixed so I use the Styd distance only for indoor. As I think about it, Stryd has basically devolved into an expensive pace stabilizer for me.
If I recommend a single accessory it is always a ECG HRM and specifically the Polar H10 with an extra strap in rotation. But for Garmin users the HRM Pro+ has enough things other than strict accuracy and durability to recommend it — including pace/distance footpod functionality.
I would love it if someone tested the HRM Pro+ footpod functionality against Stryd. And against nothing but SatIQ. I suspect the HRM-Pro+ would be slightly less good but good and the SatIQ in third position but respectable if the conditions are not too unfavorable.
he also used AW6 which is very accurate compared to modern devices.
multi frequency is NOT a substantial improvement over ambit3/v800. but it might be a small improvement.
power: yes, and power is also good for modelling load and recovery
ecg: have you looked at ECG like https://fourthfrontier.com/
HRM Pro Plus…i don’t have one sorry!
I think location matters for the GNSS stuff. The f7X is better — for instance — in Harare Zimbabwe with lots of open sky than in downtown Washington DC. Particularly seems goofy around the FBI building for some reason. ?
Chicago marathon course and Manhattan are notoriously bad for GNSS.
I have not heard of frontier. Are you meant to wear this 24×7. That seems a hard sell.
I have seen rigorous testing for the H10 and Movesense Medical along with Hexoskin and the HRM-Tri (until it proved unsuitable) over at https://muscleoxygentraining.com.
re: Chicago Marathon – https://www.wareable.com/apple/we-stress-tested-the-apple-watch-ultra-at-chicago-marathon
I use the TSS power based load and recovery in Training Peaks but it’s broken for trail running and I change it to hrTSS. Actually I stopped using Stryd a long time ago. I find hrTSS and TSS about the same for road running.
I meant to say I stopped using Stryd for specifically trail running a long time ago because they said don’t get the Stryd Wind wet and the data was questionable at best — especially if any scrambling is involved.
I’ve read/watched several reviews about fourth frontier and they were not favourable for the price over the polar h10 however it does have promise
Hi Brian, Anders here.
I understand it seems weird that I used FR310XT and V800 in those tests. I do also have a FR935 and FR645, and they were consistently less accurate in terms of distance compared to the 310XT and V800. In retrospect, I should have included a test with 935 and 645 for reference.
All GPS watches I tested provide good distance accuracy in ideal conditions. Where I run it is however a lot of trees, which makes them perform less good in the summer when there is a lot of leaves. I trust that multi-band GPS helps a lot in that regard.
What I found to be most impressive with Stryd was the spot-on instant pace accuracy as measured against the stroller wheel reference.
Unfortunately I didn’t have a multi-band watch to compare with. It would be interesting if someone could test the instant pace of a multi-band watch against for example a stroller wheel sensor. While I do trust multi-band to provide a better distance, instant pace has its own challenges.
Mutiband is a significant improvement. Under good conditions my f7X it is almost as responsive as the Stryd for pace and is more reliable for distance generally across confounding factors in terrain and shoes. There are specific urban places I can go that are consistently substantially worse than everywhere else and always have wild pace swings and a wobbly track. Deep forest like you describe might be adverse. You would have to try it.
This was a lot of work so thank you. I was definitely not doing this kind of extra projects when I had small children.
I think your methodology with the stroller is interesting. I wonder if the inertial mass of the stroller itself affects pace stability in a way that is favorable to Stryd.
It isn’t just changing shoes of the same make and model, it is changing to different shoes like speed shoes vs daily trainer vs trail shoes. I don’t know if it is the lacing placement on different shoes or wildly different stack heights. I used to use the Nike Streak LT 3, Streak 7, Pegasus 30-something, and Vaporfly 4% and found significant calibration differences with my original Stryd. Changing between Hoka Mach 4, Puma Velocity, NB Rebel v2, adidas Takumi Sen 6, Takumi Sen 7, and Boston 10 etc. also change the calibration for me with the Stryd v2 “Wind”.
I suspect the stack height is important, followed by placement on the laces, and possibly midsole material in calculating distance by dead reckoning. My mechanics versus someone else may also be important to these issues.
My sense is also that steep hills are also hard for Stryd.
I also saw some traffic a while ago on Garmin forums that the Stryd calibration is specifically wonky on Garmin watches vs Polar. There may be a Garmin problem there outside of Stryd’s control beyond the limits imposed by ConnectIQ.
The Stryd is a piece of useless device. I bought it for accuracy „out of the box“. It wasn‘t. I the meanwhile they remove this statement from their website. Even after I did a proper calibration I have issue which changing my shoes all the time. It’s easy more inaccurate than my dual frequency gps device. The Stryd team has the worst support I‘ve ever seen. They are blaming the users. In the end it’s the user experience with that device. And it’s a very bad one. I never had any issues with a distance in any race with using gps. The error was always within 1.5%. With Stryd I struggled with pace and distance all the time. If you think you are ready with a one time calibration then you are wrong.
Hi Xanders, Anders here.
I’m aware that some are struggling with Stryd accuracy. That is actually the reason I decided to test it against my stroller with a wheel speed sensor.
I’m not saying that Stryd is accurate for everyone. All I can say is that it has been accurate in my tests, measuring the pace and distance error over time during runs using two different shoes, different paces and different cadences.
I do however think that even if you have a distance inaccuracy, you will find the instant pace to be more accurate than what you get with GPS, because in my tests I found that the absolute instant pace error at a given second can be substantial with GPS watches. Whether or not that matters to you depends on how important instant pace is for you.
If you want instant pace you should use the new Coros pod 2. The pace is even more responsive than the Stryd. I have both pods.
i have both pods and coros pod is definitely not more responsive than stryd (old version) ! other reviewers say the same
And the Coros Pod 2 is Coros only…
It is indeed “useless” in the sense that it can’t be trusted. It’s bad enough that different shoes need a different calibration but I also found that with the SAME shoe the calibration drifted and after being ok for a few runs the distance measurements would vary between +/- 2% for no good reason for the next several runs. As everyone who tried to discuss these issues with Stryd found, it was like banging your head against a brick wall. Their motto is to blame any problems on their customers and refuse to even consider anything could be wrong with their device. It’s a mind-bogglingly stupid attitude, that’s for sure !
Yes, you can get better instant pace from a footpod in difficult GPS environments like urban canyons, but I use my old Garmin footpod (must have paid €30 for it back in the day) and it’s fine.
The Chicago Apple Watch vs Epix 2 test looks to me like they both performed waaaay better than my friend with a fenix 5S who had 44.52km with plenty of scribbling through buildings.
Im also Running with the stroller a lot.
There appear a few questions to me/
Did you calibrate the wheel sensor only counterclockwise on the track? This would result in ab error by itself. I don’t think it matters for training purposes, but if you are talking about fractions of a percent it should be taken into consideration.
How do you think it’s running power affected by the stroller? On my bike it’s no question because there the power can be measured, but running you have more weight to push (17kg for the Qeridoo I think plus 33 for my both kids) but with a rolling resistance. I use a Run profile on my Epix where VO2Max calculation etc is turned off because I don’t believe the algorithm for running can replicate this.
Run power can’t compensate for pushing a stroller.
Hi Bene,
The wheel sensor calibration is by no mean scientific, but should be decent. I first calibrated using a paint marker on the wheel and then doing a few revolutions on painted concrete floor in our basement, then measuring with a millimeter grade measuring tape. One revolution is approx 1.5m, and I can measure that with a few mm accuracy. A few revolutions is approx 5m which I can also measure within a few mm. That should get me below 0.3%. Then I was running the track making sure the wheel on which the sensor was mounted followed the intended path, and making sure that my indoor basement floor measurement was good.
Regarding power with a stroller; I don’t know. I use the same Thule trailer on my bike, and love that I can do power while hauling my kid 🙂 Just like you, I do a run profile without Vo2max on my FR935 and ignore power if I run hilly routes with the stroller. On flat routes I think it works ok however.
Cheers,
Anders
Unlike in cycling where the forces are measured directly with strain gauges — usually in the cranks or pedals — Stryd uses a model to estimate forward vector power based on a bunch of inputs including mass, pace, vertical oscillation, change in elevation (used to estimate grade), and wind.
It cannot measure the actual force you push against the ground. Doing things that are not in the model like pushing a stroller, pushing hands on knees, or using trekking poles will make the power estimate wrong.
This is also the source of controversy in running power as there are 2 substantially different high-level models for estimating running power and each manufacturer has a proprietary implementation of one of them — plus errors in sensor measurement on top. One model tries to estimate the total energy in the system and the other the component of energy moving you forward. Stryd and Coros is the later. Runscribe and Garmin is the former. Holding aside sensor accuracy issues, that is why Garmin power is a substantially higher absolute number than Stryd power.
Longtime Stryd user here (since 2017). Say what you will about the usefulness of power (I tend to agree with most complaints) and all those other metrics, but I have zero issues with accuracy and instant-pace. This is even more pronounced on trails with frequent switchbacks and tree cover where GPS particularly struggles.
Whatever you are testing here. My calibration factor changes depending on the used shoes from 1.9% to 4.7%. I would think gps is much more accurate than Stryd. Btw. my gps watch is mostly within 1% error on races. Stryd not
Hi Shruthi,
I read several reports online about calibration being depending on shoes. This was actually the primary reason why I did those accuracy tests with my stroller.
I did tests with two shoes, ASICS Gel-Contend and Nike Zoom Winflo. Maybe I was just lucky, or I unknowingly did an extra good job mounting Stryd in a stable and consistent way on both pair of shoes.
I find that GPS is usually within 1%, unless there is a lot of tree cover (which I suffer from, as there is a lot of forest where I live).
Mine not. I always arrived at almost the same cal factors, even when I changed the lacing method of my shoes to get rid the physical irritation from the pods ( Stryd + Tempe) to my sensitive bones) Swapping Stryd from the left shoe to the right or vice versa means a change in cal factor of not more than 0.2-0.3%, while changing the exact position of Stryd on the same shoe was always less than 1%, maybe the diff. between the two extremes were 0.7%.
Nevertheless I am not a big fun of Stryd as regards its power reading, I am quite sure it is distorted not just on steep ascents at low cadence and/or speed, but I am almost sure it underestimates power for the very robust sprints.
It is abnormal that my short term surges can not EVER result in higher very short term watts than twice my CP.
What does it mean? You can exert X watts for a very long time being close to an hour (I know that CP is not exactly CP(60min) ), but you cannot exert 3*X or 2.5*X watts not even for 5 or 10 seconds. I am not a doctor or exercise physiologist, but it is contradictory to all the other data that I know about myself like top speed (or acceleration) for 5 sec vs top speed for an hour.
I guess it comes from the approach of Stryd, they do not follow the mechanical power approach at all, but some metabolic, moreover they do some tweaks being valid for different “power ranges”. Brrrr!
Anyway I regard Stryd a pretty pretty good speedometer.