Running Power Comparisons – Apple, Garmin, Stryd, Coros & Polar
I recently had a run & bike focussed week in Cornwall, fitting in the niceties of interacting with family alongside the more important training 😉
Cornwall is very hilly and beautiful in equal measure. Many of the roads reach grades of 20% around where I was in Lostwithiel. So it was perfect territory for running power despite the oppressive heat.
It’s been quite tricky to get 5 full sets of data and there have been many problems with the collection and presentation of the data that have taken a significant amount of time to get around. It’s time to cut my losses and show you the stats of the best two Cornwall runs and then one from when I returned home.
Watches: Polar Vantage V2, Garmin Forerunner 955 Solar, Stryd (Garmin 935, handheld), I have Apple Watch 7 but used AW6 with WatchOS 9 and Coros Pace 2.
Issues & expectations: I didn’t expect to find anything different from my last published test from a few weeks back where I showed several little tests in one long run in Richmond Park, London which included some trails and small hills. However, one reader pointed out that my Apple Watch cadence figures looked wrong. I was using an unusual strap for the Aw6 and so reverted to a better one for these tests.
Lostwithiel – Eastern Loop
The route and test section show one hill and some undulation. It was hot, humid and the route was mostly sheltered from light wind.
I failed to keep any sort of constant power during each of these 5 different efforts some of which were on the flat and others on the uphill and downhill sections. Each interval was supposed to have consistent power as part of a pyramid of efforts with the end of the interval broadly coinciding with a marked change in gradient.
The cadence on the Apple Watch clearly has a problem. The Polar had some issues which were partly my fault for wearing it too loosely but also there was an incorrect timeshift in its recorded data which I had to adjust. As in the previous test, the Apple Watch fails to record power when walking – I could have sworn it DID display power this time when walking, but perhaps my memory of that fails me. (There was a WatchOS update that I had installed)
Overall power levels from Coros, Stryd and Apple are similar whereas Garmin and Polar are about 30% higher than that trio but dissimilar to each other when looked at on a mean-max chart.
Apple’s Running Power clearly shows some anomalies, notably in the 2nd of the 5 short efforts where it takes a while to get up to the ‘correct’ level.
Garmin Running Power and Polar seem to initially track each other well but then there is an unexpected widening between them toward the intervals at the end of the workout, especially in the downhill section. One of those algorithms is probably wrong for downhills.
If push came to a shove, you’d have to favour the Stryd or Coros as being the better candidates with Polar somewhat of an unknown and probably 3rd.
Penryn Loop From Lostwithiel
This test had a particularly unpleasant hill in it and the intention of the test was for a generally even power level throughout, regardless of the slope.
I can’t explain the better performance of the cadence on the Apple Watch other than to note that my cadence was fairly constant and there were no stops other than to get my bearings. That said, there were still strange Apple cadence readings perhaps coinciding with those mini stops.
Turning to the mean-max chart you might conclude that I did a reasonable job of maintaining my power level if you looked at the flat Stryd data curve. Less so if you looked at the Apple, Garmin and Coros curves.
I had set the Apple Watch to have upper and lower power alerts for this test. However, I’d set them too narrowly (or so I thought) as they were often alerting when I thought they should not and hence there seemed to be a lot of variability with Apple Running Power. You can partially see that in the final image (above) however, it was more noticeable in reality.
Anyway, that prompted me to look closer at the variability of Apple power on my return home to an even hotter and drier SW London.
3-Minute Flat Intervals, Varying Intensity – Bushy Park, London
I set the power alerts on the Apple Watch quite widely, to the point where I wouldn’t expect them to trigger…but they did. Again it’s not so visible in the data. Apple Watch lets you set a target based on either average power or instant power, so I’m thinking that the instant power is too sensitive.
Garmin and Polar seem much closer together here than in hilly Cornwall. The whole point of running power is to better handle pacing on hills…either Garmin or Polar seem to have it wrong right now. Judging by this last test I’d say it was more likely that Polar has it wrong but, then again, they don’t take into account wind and I can’t remember for this test if I had synced the Garmin Power for wind which would have made a difference today.
Apple needs to tweak their algorithm for walking pace and determine a more accurate detection of cadence.
Stryd and Coros probably have it about right with their flavour of running power.
Conclusion
They’re all different. But we already knew that.
Yeah the raw power is what it is…variable 😉 The real difference is the ecosystem and AFAIK only Stryd offers one (CP, power curves, race estimates, etc…).With the others it’s just a metric and good luck putting it to good use…maybe with Palladino’s Supercalculator but who has heard of it…
My thoughts too.
Back when I started to use Stryd in 2017 I was pointed towards the SPC, and whilst I’m numerate, I thought to myself “who’s got time for that?”
Now with PowerCentre, Auto CP and the rest of the Stryd eco-system it’s mostly automated.
The only number that most runners I know us is pace. So with that said, a live grade adjusted pace, which uses power, would be a useful standalone number.
Do you get bluetooth interference issues with so many devised used at once?
not that I’m aware of
I’ll chime in here as a cyclist first I’ve been using power in one shape or another for about 15 years. Even among power meters, there are different readings one may be high one may be low and one may be in the middle it also varies which leg is being messured or both legs. I don’t need an app to tell me my threshold power that’s an easy test or help to pace a 100-mile ride with a power meter I can do that myself. I see running power as the exact same thing as cycling it doesn’t matter which one you think reads correctly as long as you stay with one if you bounce between one or another you’re going to have issues. Stay with one train with one race with one based on all your pacing off that one and it is perfectly fine. It is the metric, not the exact number that should be your concern
yep!
platforms that continuously update CP/rFTP (like Xert, stryd and other cycling ones) are even better.
the difference with running power is that the CP curve is flatter and more sensitive around levels that you normally run at AND that the runner can more easily maintain constant power on the road than a cyclist. Even more reason to stick to one power system that appears accurate.
Nice comparison and agree that the real question is usage. Is there any running app for use with garmin, AW, polar or Coros that allows both power usage cp/ftp tests and run by power zones and also on the backend to track data and see how power ftp changes over time as a function of mileage, pace, etc. displaying power is nice but one needs to be able to use it effectively. I have styd and not crazy about the pod but the app is perfect (cp, zones, race power, history).
that sounds quite an involved analysis. id say runalyze or golden cheetah could be good places to start.