Video summary

Huawei Watch 3: A Scientific Review

Main summary

Key takeaways

Product Review

Huawei Watch 3 — Scientific review (summary of subtitles)

What the reviewer tested (scientific comparisons)

Sleep tracking accuracy (vs scientific EEG)

The reviewer compared the watch’s sleep metrics against two EEG reference devices:

  • Z-Max scientific EEG (detailed scoring of sleep stages)
  • Dream 2 consumer EEG headband (used for a shorter follow-up)

Sleep stages compared:

  • Deep
  • Light
  • REM
  • Awake

Timing checks included:

  • Sleep onset time
  • Wake time

Heart rate accuracy (vs ECG chest strap)

Heart rate was compared against the Polar H10 ECG chest strap (treated as a high-accuracy reference), across:

  • Spinning
  • Outdoor cycling / commuting
  • Longboarding
  • Weightlifting

SpO2 / oxygen saturation accuracy

SpO2 was compared against a ground-level finger pulse oximeter, and the reviewer also checked performance during:

  • Two flights

GPS accuracy

GPS accuracy was assessed by repeating the same cycling routes multiple times and checking how well tracks overlapped.

Step counting accuracy

Step counting was tested using exactly:

  • 4,000 steps total, split into four segments of 1,000
  • A mention of concern about false positives (steps counted while not actually walking) appears in the subtitles, but it is not fully confirmed there.

Key product capabilities mentioned (background)

These are feature claims, not direct performance results from the tests.

Sensors and measurements

  • Sensors:
    • Accelerometer
    • Gyroscope
    • Geomagnetic sensor
    • Optical heart rate
    • Ambient light
    • Barometric pressure
    • Temperature
  • Measures:
    • Sleep stages
    • Heart rate
    • SpO2
    • Steps
    • Skin temperature
  • Sleep algorithm:
    • Huawei TruSleep

Connectivity and extras

  • Connectivity: Wi‑Fi, GPS, NFC, Bluetooth
  • Built-in speaker / microphone
  • eSIM cellular calling (activate a plan without a physical SIM)

Water resistance and battery claims

  • Water resistant up to 5 atmospheres
  • Battery:
    • “Ultra long battery mode” up to 14 days
    • Huawei claims about 3 days normal usage (less when paired with iPhone)

Sleep tracking results — major negative

Overall conclusion from EEG comparisons

  • The reviewer says the watch is “pretty bad” at sleep tracking.
  • It is described as not reliably usable for sleep stage metrics.
  • The criticism focuses on stage classification and timing, not just totals.

Stage-by-stage performance (Z-Max EEG reference)

  • Deep sleep: ~60% correctly classified as deep
    • Often confused with light, REM, and sometimes awake
  • Light sleep: <50% correctly classified as light
    • Often confused with deep, REM, and awake
  • REM sleep: worst-performing stage
    • Only 23% correctly classified as REM
    • Most REM predicted as light
  • Awake: 43% correct
    • Awake time is frequently misclassified (often becomes light/REM)

Sleep cycle visibility / timing issues

The reviewer states that, from watch output alone, you can’t clearly see sleep cycles.

Examples of timing/staging failures mentioned:

  • Sleep detected ~30 minutes too late on one night
  • REM described as “randomly distributed” on another night
  • Extra awake periods detected when the reviewer was actually asleep
  • Awake detection described as hit and miss:
    • Sometimes correct, often wrong
    • Sometimes long awake moments falsely detected

Verification with Dream 2 EEG headband (4 nights)

The same general pattern appears:

  • Deep: 56% correct
  • Light: 40% correct
  • REM: only ~25% (about 1/4) correct
  • Awake detection: slightly better than the Z-Max comparison, but still mis-detects extra awake time
  • A repeated issue remains: falling asleep detected too late, including one case over an hour late

Comparison vs other sleep trackers (context)

The reviewer compares Huawei Watch 3 against devices described as better performers:

  • WHOOP
  • FIBO Inspire 2
  • Withings Sleep Analyzer (under-mattress)

Key claim:

  • These were said to perform better for all stages, especially REM.

Sleep onset / wake time accuracy (vs EEG)

  • Watch predictions were compared to EEG times (minute/hour differences).

The reviewer reports:

  • Wake time is often close, but with large outliers:
    • Woke up ~30 minutes early once
    • Fell asleep ~30 minutes late another time
  • With Dream 2 reference:
    • One night had falling asleep more than an hour late

Overall:

  • Mostly close, but extreme errors occur.

Heart rate accuracy — mixed, depends heavily on activity

Reference standard

  • Polar H10 ECG chest strap as ground truth.

Spinning (generally good)

  • Measurements mostly match closely.
  • Deviations exist, but described as not bad overall.

Outdoor cycling / commuting (weaker)

  • Watch follows broad trends but misses peaks and dips.
  • Possible contributors suggested:
    • Bumps/movement
    • Sweat
    • Movement effects beyond true HR changes

Longboarding (better than cycling)

  • Less motion is expected, and results are described as better:
    • More peaks/dips match than with cycling.

Weightlifting (struggles)

The reviewer attributes errors to wrist/flexing effects during lifting.

Reported results:

  • Higher heart-rate peaks are often detected too low
  • Watch can follow trends but misses peaks of each set

Heart-rate verdict

  • Overall described as “okay”:
    • Good: spinning, longboarding
    • Struggled: cycling, weightlifting
  • Recommendation:
    • For more reliable HR during exercise: Polar H10 chest strap

SpO2 / oxygen saturation — sometimes works, not consistent

Ground-level comparison (78 measurements)

  • Values generally match the finger oximeter range.
  • Mostly normal readings, with one noted lower-ish measurement:
    • One instance around 93%

During flights

Expected behavior:

  • During ascent and cabin pressure changes, SpO2 should generally show a drop and then recovery.

Observed results:

  • Huawei Watch 3 followed the expected pattern sometimes
  • But results were mixed:
    • Sometimes too low
    • Sometimes not low enough (higher than expected when it should be lower)

Overall claim:

  • It can sometimes detect significant drops, but is far from perfect.

SpO2 verdict

  • Not terrible, but inconsistent.
  • The reviewer also notes other devices perform better (no specific competitor named in the subtitles for SpO2).

GPS accuracy — mixed but “okay”

Test method

  • Same cycling route repeated 5 times outward and 5 times return.

Findings

  • Starting points usually cluster well (good acquisition)
  • Otherwise tracks are somewhat close but not perfectly road-following:
    • Significant deviations through buildings occur
    • Some weird artifacts occur
  • Return tracking was described as better:
    • Closer overlaps and better road adherence

Data export note

  • Watch data couldn’t be exported in a common format.
  • Reviewer converted data to GPX to use in Google Maps.

GPS verdict

  • Somewhat mixed confidence:
    • Acquisition can be slow
    • Accuracy can drift noticeably

Step counting — generally good (but overcount tendency)

Method

  • Exactly 4,000 steps total in four segments of 1,000
  • Counted with a telecounter
  • Worn on the right arm; telecounter position switched as needed

Results

  • Watch readings were close to the 1,000-step segments
  • Slight tendency to overcount, not dramatically

Comparison vs Galaxy Watch 4

  • Galaxy Watch 4 tends to undercount
  • Huawei Watch 3 tends to overcount
  • Overall both were said to be about equally well, with Huawei possibly slightly better

Step counting verdict

  • “Pretty good” for walking counts
  • Reviewer notes they still needed to check false positives (steps while not walking, e.g., cycling/weightlifting), but subtitles stop before confirming results.

Overall pros / cons (as stated in the review)

Pros

  • Looks / UX: reviewer calls it “beautiful” and likes the interface
  • Step counting: accurate enough for walking; only slight overcount
  • Heart rate: acceptable for some activities (spinning, longboarding)
  • SpO2: can sometimes detect lower saturation
  • GPS: described as “pretty okay,” though inconsistent

Cons (biggest emphasis)

  • Sleep tracking is very poor, especially:
    • Wrong sleep-stage classification (notably REM)
    • Extra “awake” periods
    • Unreliable sleep cycles
    • Sleep onset time off by 30 minutes to over an hour
  • Heart rate struggles with cycling and weightlifting (missing peaks)
  • SpO2 during flights is inconsistent (sometimes not low enough)
  • GPS not consistently road-accurate (sometimes deviates into buildings)
  • Price is criticized (“not cheap”)

Comparisons / recommendations made in the subtitles

  • Better-value alternative claimed:
    • Me Watch: about one-third the price and “performs better overall” (details not fully provided in subtitles)
  • For heart rate:
    • Prefer Polar H10 chest strap
    • If on iPhone, reviewer suggests Apple Watch for more accurate HR (in their view)
  • For sleep tracking:
    • Prefer Fitbit devices (examples mentioned):
      • Fitbit Luxe / Charge 5 / Inspire 2 (subtitles include “Vivid Lux… Vivid Charge 5… Inspire 2,” likely referring to these models)
    • Under-mattress:
      • Withings Sleep Analyzer performs reasonably well
    • WHOOP:
      • Good HR and sleep tracking, but subscription increases long-term cost

Speakers / contributors

  • Single speaker (“Rob”): the evaluation appears to come from one reviewer, with no other contributors.

Concise verdict / recommendation

Do not buy the Huawei Watch 3 if sleep tracking or reliable health metrics are your priority.

  • Sleep stage tracking is described as severely inaccurate (especially REM).
  • Heart rate performance is activity-dependent (good for some workouts, poor for cycling/weightlifting).
  • GPS and SpO2 are mixed/inconsistent.
  • Step counting is reasonably good, but not enough to recommend the watch at its price.

Original video