Apple Watch Running Power Data Comparison to Garmin, Stryd, Polar & Coros

This article compares running power data from all the major running watch companies.

Q: What Is Running Power?

Running power is a measure of effort, although physicists might prefer the term ‘work’.

If you run at a constant speed on flat ground and then maintain that speed up a hill, your running power will increase up the hill simply because it’s harder. Your heart rate will also increase with a lagged effect.

Q: How is running power measured?

Running power (effort) is affected by several factors including body weight, wind, slope & ground conditions. It is calculated (measured) by one of two methods – metabolic cost or Gravity/Velocity (GOVSS) – using sensor data such as accelerometers, GPS and barometers.

This sensor data can be from inside a watch or come from external sources such as a Stryd pod or weather (wind) forecast. It’s perhaps a more elegant solution to have everything determined inside the watch but not necessarily a more accurate solution. That said, there is no accepted definition of ‘correct’ like there is for cycling power.

Running Power – A Very Brief History

The Stryd pod accurately measures distances and Stryd was the company that started the running power market. Every major running watch company will let you use the Stryd pod as a source of running power but Garmin, Polar, Coros and now Apple also have their own calculations for running power meaning that you do not need Stryd but the results will be different to those from Stryd.

Apple Running Power is currently available to Watch Series 4 and above via a public beta. I’ve tested that here against Stryd, Polar, Coros and Garmin Running Power to get a feel of how it performs.

The Tests & Equipment Used

I used the following watches

• Apple Watch with watchOS 9 beta on AW6 (I also have AW7, AW6 appears more accurate). I used a 12South sweatband to hold the Apple Watch.
• Garmin Forerunner 955 which produces Garmin Running Power on the wrist in conjunction with a 2nd device Garmin HRM-TRI in this case. I accidentally kept Stryd as a source of distance/pace for the 955 which should increase accuracy.
• Coros Pace 2 (nothing else required, Coros has an optional pod)
• Stryd Pod (latest generation wind pod) with data displayed on a Garmin Forerunner 935. I had this on Hoka Rincon 2 shoes with normal lacing.
• Polar Vantage V2.

Test Conditions: A very hot day with no wind. All test segments of the run were in Richmond Park, London and were mostly on heavily compacted dirt or hard path (generally non-tarmac).

I had all the watches on my arms. This is far from ideal but I only have 2 arms and the watches need to ‘swing’ to produce meaningful results.

My weight was set correctly in each app and synced where appropriate. The only source of error here is that Stryd only let me enter whole Kg values and the value I entered was 300g too low.

I used HealthFit to sync the data from the Apple Watch as RunGap doesn’t yet support running power.

I used another tool to get a separate, ‘correct’ elevation track.

The data is shown with DCR’s Analyzer software but I had to manipulate one of the source files so that all 5 sources of power could appear on one graph (#ItsComplicated). Generally, I will show charts with 10-second smoothing and if you want to look more closely for yourself check out the following link.

Link To: DCR Analyzer Data Set, Charts & Analysis

I performed a series of small tests to see the effects of running/walking uphill, downhill and on the flat. At the end of each mini-test, I stood still to make it easy to distinguish each part.

some Obvious Concerns

If we check the raw data for cadence then clearly there is something amiss with the Apple Watch. If it can’t get the number of steps to be correct then you do have to doubt the accuracy of the rest of the piece.

 

The elevation tracks do not seem to match the ‘corrected’ track in yellow. Again the Apple elevation seems to go under or over the corrected track whereas the other brands seem to track it fairly consistently.

The track labelled Stryd is the one from the Garmin 935 and that won’t be used by the Stryd pod.

 

Turning to the overview power-duration curves we can see that the 5 sets of results are all quite different although Coros and Stryd are scarily similar. Apple is also vaguely similar to Stryd. I suspect that Apple will refine their algorithm more closely toward Stryd’s results over time.

Polar and Garmin are fairly similar to each other and it’s not an issue of their inaccuracy, rather, a different method is used.

 

 

10 Tests, 1 Run

There were 10 tests, numbered on the chart above

1. ‘Hard’ walk off road up a steep hill
2. Easy walk off-road on a relatively flat section.
3. Off-road jog and then hard spike up some steps
4. A km or so running on an easy downhill at 4:00/km ish.
5. Attempting to maintain a constant power going uphill for a couple of minutes with a hill that progressively steepens
6. Running more quickly downhill on-road
7. Another steepening hill culminating in steps to the top
8. Twisting, undulating downhill
9. Walk
10. slightly downhill walk

 

Tests 1, 2, 9 & 10

The walking tests gave strange results from Apple. Test “1” has Apple Watch reporting over 200w. Fair enough it was steep. However, subsequent walks on tests 2, 9 and 10 report no power which is incorrect. At some indeterminate point when walking Apple decides to produce a power figure…or not

Test 3

A relatively constant effort should have been recorded followed by a sharp spike in the effort at the finish. To me, the chart shows that all the watches except Coros handle this plausibly. Coros seems perhaps a little laggy and muted.

Test 4

Again Coros seems laggy in this downhill run. Conversely, Garmin seems too responsive and spikey at the start and I can’t remember exactly how I finished the section to comment on the unusual Garmin peak at the end.

Test 5

I thought I did a fairly good job of maintaining a mostly constant power running up this hill even though the gradient steepened more towards the top. Polar and, to a lesser extent, Garmin seems quite variable compared to the reality I remember and which the other three watches report.

Test 6

I endeavoured to maintain a constant pace down this hill on the road but I know that I failed to in reality. My memory of it is a smoother start as shown by Apple but with an increased effort as shown by Stryd from there on to the end.

Test 7

this is a short off-road hill which steepens considerably toward the end at which point there are steps. Apart from Coros, all the other 4 give a plausible result here. Normally on this hill, I would run at maximum power toward the apex but this time I tried to maintain a lower and more even output.

Test 8

This downhill trail was tricky at the start as there were brambles, bends and a small gully to contend with. Consequently, I would expect to see variability in the level of the power track but not the peak that Garmin shows nor the lag that Coros shows.

 

Take Out

N=1 and the results from Apple appear plausible.

This is only beta software from Apple and I’ve only done this one full-blow test against all the other watch systems. I have a few other runs worth of data but nothing firm enough to draw any conclusions as each time there was only one other watch in play.

What stands out in my mind for this test are three things

1. Apple power is incorrect when walking
2. Coros is not sufficiently responsive to power level changes
3. Garmin seems to be too sensitive sometimes and gives unwarranted spikes.

There are, of course, a thousand and one other possible tests that can be done and I see that dcrainmaker has some test results for runs on road-like surfaces (links to his post), so that’s a good link to see results on different surfaces. I would also be intrigued to know if and how Apple accounts for wind and how it handles more extreme ground conditions like snow or sand, it’s the extremes where even Stryd falls over.

Apple has introduced several notable and new sports features including new triathlon functionality. There are also other running power-related features that I’ll cover later in the year when this is all properly released.

Apple’s beta software that’s used here is probably very, very close to how it will be fully released in August/September. However I like what I see, Garmin, Polar and Coros should be worried about new competition for runners with the range of new Apple features whereas Stryd might well be worried that there is now an Apple-only alternative for what was probably one of their more lucrative platforms.

I have zero intention of switching from Stryd. Once you switch platforms your historic data will not be valid.

Must Read: Stryd Review

 

Apple Triathlon – Good First Time Go At Triathlon On the Watch

See more