Compare How I checked

How I checked Stamio

Since May 9, 2026, I've run with two watches: my Apple Watch Ultra and a Garmin Fenix 8. Every Stamio calculation was compared, day after day, with what the Fenix showed. Here is the method, the gaps, and what doesn't match yet.

109runs compared

89nights compared

5 monthssince May 9, 2026

The method

1
Two watches at once, since May 9
The Apple Watch Ultra for Stamio, the Garmin Fenix 8 as the reference, worn together at night and on runs.
2
The same heart rate strap since October 22
A Garmin HRM 600 chest strap sends its heartbeats over Bluetooth to both watches at once. They receive exactly the same heart rate: a gap can no longer come from the sensor, only from the math. Before that, Stamio read the wrist or AirPods Pro.
3
The Fenix is only for comparison
Stamio calculates everything from the Apple Watch's measurements. No Garmin value goes into the app.
4
Months, not a day
A tuning is kept only if it reduces the gap over the whole history, leaving out one day at a time to check it doesn't fit by chance.

The results

What Stamio calculates	Average gap with the Fenix	Compared over	Verdict
VO2 max	0.28 ml/kg/min	109 runs, June → October	Very close
Race predictions	under 0.7%	5K to marathon	Very close
Overnight resting HR	under 1 bpm	every night since August	Very close
Aerobic training effect	0.2 out of 5	59 runs	Close
Anaerobic training effect	0.3 out of 5	59 runs	Close
Overnight HRV	4.8 ms	93 nights	Close
Sleep score	4.9 points out of 100	89 nights	Close
Morning readiness	5 points out of 100	September and October	Close
Energy	7 points out of 100	every 3 minutes, watch worn	Close
Recovery time	4 hours	in October	Fair
Load of a run	11%	59 runs	To improve
Acute load (7 days)	10%	42 days	To improve

Average gap: the difference between Stamio and the Fenix, ignoring the sign, on average. Apple Watch Ultra measurements against the Garmin Fenix 8, May to October 2026.

Since October 22: same strap, run by run

Both watches receive the same heartbeats. Average heart rate is identical; what's left is the calculation gap.

Fenix 8Stamio

Workout	Distance	Avg HR	VO2 max	Aerobic effect	Anaerobic effect	Load
10/23 · easy run with strides	7.35 mi7.31 mi	148148	51.651.9	3.12.9	1.81.2	11190
10/25 · long run	9.02 mi9.03 mi	152152	51.551.7	3.33.2	0.00.0	9995
10/27 · easy run	6.26 mi6.24 mi	148148	51.451.6	3.02.8	0.00.0	7569
10/28 · 4 × 1,000 m	5.16 mi5.14 mi	161161	51.952.0	3.63.5	3.02.7	210193

Fenix VO2 max: the day's precise value in Garmin Connect (the watch shows the whole number). Strength session on October 22: load of 59 on both.

What to take away

Distance within 0.5%, identical average heart rate: both watches saw the same workout.
VO2 max within 0.1 to 0.3, aerobic effect within 0.1 to 0.2.
Load stays 4 to 19% below the Fenix, mostly on workouts with surges: that's the next tuning.

The night: same heart, not quite the same night

For HRV and sleep, the gap hardly comes from the measurement anymore. It comes from two choices each watch makes: when it decides you're asleep, and which stages it thinks it sees.

Overnight HRV

12.4 → 4.8 msaverage gap with the Fenix, before and after calculating from heartbeats (93 nights)

2.9 msmedian gap on well-measured nights, since August (50 nights, correlation 0.96)

7 mshow much my HRV changes from one night to the next, median, according to the Fenix

Stamio calculates HRV from the heartbeats themselves (RMSSD), like sports watches, instead of taking the Health app's value (SDNN). Both watches then measure the same thing: the gap that's left is smaller than what changes from one night to the next.

Part of it comes from what each one calls “the night”: it decides on its own when you fell asleep and when you woke up. When both agree within 15 minutes, the average gap is 3.2 ms. When they don't, one night in three and mostly at wake-up, it rises to 4.2 ms.

They don't sample the same way either: the Fenix measures continuously, the Apple Watch in one-minute series, about thirty a night since August. Before, it took only 3 to 5: those are the least precise nights.

Sleep stages

0.53 / 0.21agreement with the sleep lab (kappa, out of 1): Apple Watch Series 8 / Garmin Vivosmart 4, study published in 2025

−25 / +44 minof deep sleep a night: underestimated by the Apple Watch, overestimated by the Garmin, in the same study

43 / 100 minof deep sleep a night on average: my Apple Watch / my Fenix 8 (89 nights)

Stages are the weakest point of watches. A study published in 2025 compared six of them with a sleep lab recording (polysomnography) in 62 adults: the Apple Watch Series 8 agreed best, the Garmin tested the least. It was a Vivosmart 4, not a Fenix 8, but on my nights the gap goes the same way: the Fenix counts almost an hour more deep sleep a night.

So Stamio keeps the Apple Watch's stages. That choice matters: they feed the sleep score, and the sleep score feeds morning readiness. The score follows the structure of the Fenix's (duration, stress, awakenings, share of deep and REM sleep) and stays within 4.9 points of it on average, without borrowing its stages.

Night window: from falling asleep to waking up as seen by each watch. Gaps calculated on the nights both watches were worn, April to October 2026.

Where Stamio differs on purpose

The night as the Apple Watch sees it

Stamio takes the night as your watch sees it, with a 30-minute margin on each side so nothing of falling asleep is lost. Aligning the window with the Fenix's didn't reduce the gap (5.2 ms vs 4.9): I didn't do it.

The Apple Watch's stages

Rather than imitating the Fenix, Stamio keeps your watch's stages, the ones closest to the sleep lab in the 2025 study.

VO2 max to the tenth

The Fenix shows a whole number; Stamio shows one decimal. So you see 0.1 or 0.2 changes that the Fenix rounds away: it isn't more noise, it's the same number, seen up close.

What about the Health app's VO2 max?

Stamio never uses it: it calculates its own from your runs (heart rate and speed).
A study published in 2025 compared the Apple Watch's VO2 max with a lab exercise test in 30 people: it was underestimated by 6.07 ml/kg/min on average, with a 13.3% average error.

The limits, plainly

One runner so far: me. A man who runs 3 to 5 times a week, from 10K to the half marathon.
The reference is a GPS watch, not a lab. It makes estimates too.
Load often runs a little low, and readiness a little high in midsummer: known, to be fixed.
The Apple Watch isn't perfect either: it underestimates deep sleep (by 25 minutes in the 2025 study). Stamio's score is tuned on its own stages, so it doesn't penalize you wrongly.
In the first weeks, VO2 max settles in, as on a new watch.

Next: the beta

Testers who also run with another watch will be able to share an export of their values, with their explicit consent. The table above will become one of several runners, published here.

Questions

Why compare with a Garmin?

Because it's the watch many runners want to leave without losing their landmarks. If Stamio gives the same orders of magnitude, you don't change your mind about your fitness by changing watches.

Will these numbers hold for me?

The calculations aren't tuned on me: they start from relative values (your resting heart rate, your max HR, your history). The beta is there precisely to check them on other profiles.

Can I redo the comparison?

Yes: wear your GPS watch and your Apple Watch together for a few weeks and compare. Ideally with one chest strap linked to both.

Sources

Lambe R., O'Grady B., Baldwin M., Doherty C. Investigating the accuracy of Apple Watch VO2 max measurements: A validation study. PLoS ONE 20(5): e0323741, 2025
Schyvens A.-M. et al. A performance validation of six commercial wrist-worn wearable sleep-tracking devices for sleep stage scoring compared to polysomnography. Sleep Advances 6(2): zpaf021, 2025
Fenix 8 values: my Garmin Connect account; Stamio values: the app on my iPhone (read on October 29, 2026).

Try it on your watch.

The beta is free, no account. A year of Stamio Pro for every tester at launch.

Join the beta See pricing

Other comparisons

Estimates for wellness and training, not a medical device.