Measurement log
On flat ground, Garmin's running power is just your pace
26 runs from my own export, plotted. Pace alone explains 94.5% of what the watts do. What that means before you buy a power meter.
Garmin puts a watt number on every run now, no extra sensor required. It sits next to pace and heart rate as though it were a third, independent thing you could train by.
I exported 26 of my runs and plotted the watts against the pace. They are the same measurement.
The plot
Source: my own Garmin Connect account export, summarizedActivities, all 26 running activities between 17 June and 21 August 2026. Line is a least-squares fit of power on pace.
Every run sits on the line. A least-squares fit using pace and nothing else explains 94.5% of the variation in the power figure. The typical run lands within 8 watts of what pace alone predicts, across a range from 316 to 438 watts.
For a flat-road runner, that is the whole story. The watts are your pace, multiplied by about −62 and shifted. Watching them tells you what watching your pace already told you.
Why this is not a scandal
Garmin’s wrist-based figure is a model, not a measurement. There is no force sensor on your wrist. The watch takes pace, cadence, vertical oscillation and the barometric altimeter and estimates the mechanical work. On flat ground, with steady cadence, pace dominates that estimate — so of course the output tracks pace. The model is behaving exactly as designed.
The claim being made for running power is a different one: that on hills and in wind it captures effort that pace hides. Uphill your pace collapses while your effort does not; power is supposed to stay honest. That is a reasonable claim, and my data cannot test it.
| What this data can say | What it cannot |
|---|---|
| Wrist power adds nothing over pace on flat roads | Whether power beats pace on hills |
| 94.5% of the variance is pace, across 26 runs | Whether a Stryd foot pod behaves differently |
| The residual scatter is about 8 W | Whether power helps in strong wind |
What follows for you
If you run on flat roads, ignore the number. It is not wrong, it is redundant. One fewer metric on the watch face is a small gift.
If you run hills, this data says nothing about your case — and be careful with anyone who tells you otherwise from a flat dataset like mine.
And before buying a dedicated power meter: what I have shown is that Garmin’s wrist estimate is pace-derived on flat ground. That is not evidence that a Stryd is useless, and it is not evidence that it is worth £200 either. A foot pod measures differently, and I do not own one, so the box at the top of this page lists it under specification comparison, not tested.
If someone sells you running power as a new training signal, the honest question is: on your terrain, does it move independently of your pace? For me it does not. That took one export and one plot, and you can run the same check on your own data in twenty minutes.
How to check this yourself
Garmin Connect → Account Settings → Export Your Data. In the archive, open
DI_CONNECT/DI-Connect-Fitness/…_summarizedActivities.json and pull avgPower, distance and
duration for your runs. Divide duration by distance for pace, then plot pace against power.
If your points scatter instead of forming a line, you run more varied terrain than I do — and for you the metric may genuinely be adding something.
The honest limitations
One runner, one watch, one region of northern Germany, 26 runs, all on roughly flat ground. The relationship I found is strong enough that terrain is the only plausible escape from it, and that is precisely the escape I cannot rule out. Treat this as a finding about flat-road wrist power, not about running power as a concept.
Sources
- My own Garmin Connect account export:
summarizedActivities, 26 running activities between 17 June and 21 August 2026, fieldsavgPower,distance,duration,elevationGain.