Heart rate and heart-rate variability

Updated 28 Oct 2026

Role: Trend · Current decision: Match the measurement window

Distinguish the measurements

Heart rate is beats per minute. HRV describes variation between successive beat intervals; its numerical value depends on the statistic and sampling window. A minimum HR, overnight average HR, predicted waking HR and HRV are separate quantities.

Oura’s overnight HRV display averages five-minute samples. COROS describes averaging ten-minute intervals. COROS’s current resting-HR estimate predicts seated HR shortly after waking from sleeping measurements. It is not simply the lowest sleeping value. Oura HRV, COROS HRV, COROS resting-HR definition.

The Oura API also distinguishes its sleep-HR average and minimum from the app display: these API values use thirty-second samples, whereas the app uses aggregated five-minute samples. A small API/app difference can therefore arise without a sensor disagreement. API reference.

The first observation

Oura returned 50.4 bpm average during sleep, a 45 bpm sleep minimum, and 61 ms average HRV. COROS returned 61 ms overnight HRV, 44 bpm daily minimum, 64 bpm daily average and 55 bpm predicted seated resting HR. October 13 record.

The identical rounded HRV values are encouraging agreement for that summary. They do not verify identical beat intervals or processing. The 44 versus 45 minima also cover different scopes. The 55 versus 45 difference is not a demonstrated 10 bpm sensor error, because the definitions differ.

COROS recorded average HR 148 during the previous evening’s track run. There were no returned Oura samples in that workout interval, so that session does not support an exercise accuracy comparison.

The evening decline is a different measurement

Before the sleep ending 15 October, I noticed a decline from the high 70s around 22:00 to the high 50s around 01:30. The targeted API query broadly supports a declining sampled pulse: awake-labelled half-hour medians were 82 bpm at 22:00, 63 at 23:00, 56 at 00:30 and 55 at 01:00. Values fluctuated and coverage was irregular. The record shows sample counts, ranges and gaps; differently sampled workout/rest categories were excluded from this awake-only comparison.

This is not a continuous resting-heart-rate trajectory or a test that isolates dinner. Food, posture, movement and settling can coincide with clock time. A randomized meal/autonomic experiment supports a post-meal cardiac response, but does not identify the cause or duration of this personal trend. The study had Flax Council of Canada and NIH support; West reported consulting for the Flax Council. Sauder et al..

The overnight mean was 49.375 bpm, compared with 49.5 the previous night; the minima were 47 and 45. Those summaries are separate from the evening readings. Earlier meals can be tested for sleep, comfort and functioning without making the lowest possible bedtime pulse the success criterion.

Before the 16 October sleep, I reported pulse in the high 40s, the lowest personally recalled while awake. After synchronization the focused Oura query returned 47 bpm at 23:53:59 and 48 bpm at 23:59:15, both labelled awake. These samples support the reported range, without establishing a continuous resting-HR curve or an independently verified lifetime minimum. The half-hour sample medians declined from 60.5 to 56 bpm, with irregular coverage. Overnight mean HR was 49.125 bpm and minimum 45; average HRV was 51 ms versus 58 the preceding night. The later COROS retrieval returned 56 ms official sleep HRV, compared with 55 the preceding night. Its daily snapshot reported average HR 52 bpm and minimum 43, while predicted seated resting HR was 51; those scopes do not match Oura’s sleep-window HR. These scopes remain distinct; a meal effect has not been isolated.

The social-night exception and a provider-returned range

On 17 October, after the reported hard workout, late food, vodka and uncomfortable floor sleep, Oura returned mean sleeping HR 63.625 bpm and minimum 56, versus 49.125 and 45 the preceding night. Its sleep HRV fell 51 → 28 ms. COROS official sleep HRV fell 56 → 32 ms and was labeled Below normal, with an explicitly returned 42–64 ms normal range and 53 ms baseline. These are COROS’s own assessment values, not thresholds reconstructed from our few earlier observations. Its normal range comes from previous nights and is not a diagnostic boundary or a range to apply to Oura. COROS definition.

The cardiac change accompanies a shorter, subjectively uncomfortable night. Training, meals, alcohol, timing and sleeping conditions changed together; their isolated contributions remain unresolved. COROS’s daily HR snapshot of 71 bpm average and 52 minimum, with predicted seated resting HR of 59, retains its different scope. A return toward the familiar series over subsequent ordinary nights would be informative; this one record does not establish an arrhythmia or a new long-term baseline.

The following long night

On 18 October, Oura’s sleep-HR mean returned from 63.625 to 51.625 bpm, with minimum 46 versus 56. Its average sleep HRV rose 28 → 59 ms; COROS’s official sleep HRV rose 32 → 61 ms, now labeled Normal with a returned 44–64 ms range and 54 ms baseline. Each day retains its own provider-supplied range; the earlier 42–64 range and baseline 53 remain in the October 17 record. A change in provider reference values does not justify silently rewriting history.

This follows a much longer sleep episode, after food around 20:00, gym ending 22:00 and a midnight sleep attempt. It is a return toward earlier readings across one subsequent night, not proof that one exposure caused the change or that full recovery is complete. COROS’s daily-average HR snapshot was 53 bpm, daily minimum 45, and predicted seated resting HR 55; these remain separate from Oura’s selected sleep-window mean and minimum. The October 18 follow-up reports feeling awake on waking; later-day experience remains unreported. It also retrospectively describes October 17 as feeling good despite that night’s lower HRV and grades.

October 19: lower sleeping pulse, higher HRV

The October 19 record returned Oura mean sleeping HR 47.625 bpm, minimum 44, and average sleep HRV 69 ms. COROS official sleep HRV was 67 ms, labeled Above normal against its returned 46–66 ms range and 56 ms baseline. Its daily snapshot averaged 53 bpm with minimum 42, while predicted seated resting HR was 53. Each retains its own scope and provider assessment; the COROS label is not a diagnosis or a range to apply to Oura.

This follows reported stressful activity before bed, with sleep described as okay and multiple brief remembered awakenings. Those conditions do not contradict the measurements, but their isolated effects are unresolved. The ring configuration now reports firmware 2.1.36, previously 2.1.25, with the sleep algorithm field still v2. Exact update timing and any effect on the estimates are unknown, so the version boundary remains part of the interpretation. The October 19 follow-up reports feeling pretty sleepy initially and back to normal speed within 10–15 minutes; later-day experience remains unreported.

October 20: cardiac summaries and the experienced night

On October 20, Oura mean sleeping HR was 47 bpm, minimum 43, and average sleep HRV 70 ms. COROS official sleep HRV was 71 ms, labeled Above normal against its returned 47–69 ms range and 58 ms baseline. Its daily snapshot averaged 52 bpm with minimum 42; predicted seated resting HR was 49. These scopes remain distinct. Firmware 2.1.36 and sleep algorithm v2 match the preceding retrieval.

The diary describes okay-ish sleep, many brief remembered awakenings and plentiful dreams. High HRV and a lower sleeping pulse do not erase that experience or establish complete recovery. The follow-up reports waking around 09:30 feeling super sleepy, then an unsuccessful 5–10-minute attempt to sleep more. A second follow-up clarifies feeling normal within 10–15 minutes; sustained later-day function remains unreported. No snoring or breathing symptoms were known to the user, which is not a negative clinical test. The reported roughly 5½-hour meal gap and unquantified later sips have not been isolated as explanations for the measurements.

October 21: fewer remembered awakenings, different cardiac summaries

On October 21, Oura returned mean sleeping HR 50.125 bpm, minimum 46, and HRV 62 ms. COROS official sleep HRV was 63 ms, Normal, against its returned 48–68 ms range and 58 ms baseline. Its daily HR snapshot averaged 52 bpm with minimum 44; predicted seated resting HR was 52. These are distinct scopes. HRV fell from 70/71 ms while the diary reports fewer awakenings and sleepiness clearing within 10–15 minutes. That combination does not establish a loss of recovery or make either metric a verdict on experienced sleep.

October 22: interrupted recording limits comparison

The October 22 record has COROS official HRV 60 ms, Normal, range 49–69, baseline 59. Oura HRV 63 ms, sleeping mean HR 49.125 bpm and minimum 45 describe only its recorded segment ending 06:30, after reported low battery. They are not full-night matched-window comparisons with the watch. COROS daily HR averaged 54 bpm with minimum 44 at retrieval, and predicted seated resting HR was 56; these scopes remain distinct.

October 23: full main episode and matching HRV summaries

On October 23, Oura mean sleeping HR was 49.125 bpm, minimum 46, and HRV 62 ms. COROS official sleep HRV was also 62 ms, Normal, against its returned 49–69 ms range and 59 ms baseline. Both main windows extend to about 09:28; identical summary HRV does not validate beat-level agreement. COROS daily HR averaged 54 bpm with minimum 45 at retrieval; predicted seated resting HR was 52. Today’s Oura values cover a longer window than the known incomplete October22 recording, so their similarity alone is not a matched-window comparison.

October 24–25: a change needing context

The October 24 and October 25 retrievals returned Oura average sleeping HR 49.875 and 65.75 bpm, minimum 45 and 59, HRV 51 and 37 ms, and peripheral-temperature deviations +0.15 and +1.34 °C. Readiness was 90 and 42. The summaries moved together but do not isolate one mechanism, measure core temperature or establish illness severity. The follow-up specifically reports sickness, alcohol, late bedtime and pre-bed emotions for October 25. Their separate effects are unresolved. A later COROS retrieval adds official sleep HRV 55 ms Normal on October 24 (range 49–69, baseline 59) and 40 ms Low on October 25 (range 50–68, baseline 59). Both HRV series fall, but this does not isolate a cause or validate exact agreement. COROS daily HR snapshots averaged 66 and 73 bpm, minima 43 and 53; predicted seated resting HR was 54 and 61. Those scopes differ from Oura’s sleeping means. No individual beat comparison is inferred.

October 26: main-night readings move toward earlier values

On October 26, Oura average sleeping HR was 53.25 bpm, minimum 48, HRV 53 ms and peripheral-temperature deviation +0.01 °C. COROS official sleep HRV was 51 ms Normal, range 47–67, baseline 57. These move toward earlier readings after October 25’s changed summaries. The follow-up nevertheless reports ongoing body pains, bad sleep, waking sleepy and feeling unhappy overall. Near-baseline temperature deviation and improved HRV do not override those symptoms or establish performance recovery. COROS daily HR averaged 56 bpm, minimum 45; predicted seated resting HR was 56. The Oura evening rest episode has its own HR/HRV and is not merged into the main sleep.

How much confidence is justified?

October 28: Oura mean sleeping HR 51.75 bpm, minimum 46, HRV 55 ms, Readiness 89, temperature deviation −0.02 °C. COROS official HRV is 54 ms Normal, range 47–67, baseline 57. HRV fell from 64/63 ms without establishing poorer experienced recovery. The follow-up reports a good morning despite lower HRV, and a presleep impression that pulse had not settled as desired. Those are separate scopes from nightly mean HR/HRV; the presleep impression is not a matched pulse comparison or proof of a meal/light effect. Current body pains and sustained later-day functioning remain unspecified. COROS daily HR snapshot is 55 bpm average / 45 minimum, predicted seated resting HR 52, separate from sleep-window HR. Record and curves.

October 27: Oura mean sleeping HR 47.875 bpm, minimum 44, HRV 64 ms, Readiness 94, temperature deviation −0.12 °C. COROS official sleep HRV is 63 ms Normal, range 47–67, baseline 57. Its daily HR snapshot is 48 bpm average / 41 minimum, predicted seated resting HR 49, separate from Oura’s sleep scope. The follow-up independently reports feeling good during the night and in the morning, with almost all body pains gone. This supplies experienced improvement alongside the measurements, while complete symptom resolution and sustained later-day functioning remain unestablished. Record and curves.

Independent 2025 work compared Gen3 and Ring 4 with a Polar H10 electrical chest strap. Thirteen people contributed 536 nights overall; Ring 4 had 138 usable RHR nights. Mean absolute percentage error was 1.94% for nocturnal RHR and 5.96% for HRV, with considerable variation between nights. AFRL funded it; authors declared no competing interests. This supports those tested nightly summaries, not Ring 5 or PACE 4 accuracy. Dial and colleagues.

The current product’s “99%” HR headline is an r-squared claim citing older research. The original 2020 paper describes second-generation hardware and 49 analyzed nights, with Oura employment and equipment support disclosed. Correlation is not the fraction of readings within a chosen bpm error. Original paper, product footnote.

A separate Gen3 investigation found HRV agreement depended on signal-quality filtering and improved with longer aggregation. Aggressive filtering also discarded data. A clean-looking average can therefore conceal missing periods. Signal-quality study.

The target

Develop a repeatable personal series, then examine changes in relation to training, sleep, illness and functioning. We have not selected a personal HRV maximum or HR minimum as the sleep goal. A change earns its interpretation from the circumstances and relevant outcomes, not its direction alone.

For comparisons, specify the HRV statistic, interval length, selected sleep episode, valid coverage, artifact handling and aggregation. Keep each device as a named series. Do not splice the highest HRV or lowest HR into a composite result.

Reference choices

A validated electrical chest strap offers a practical comparison for HR and beat intervals; ECG provides a clinical or research reference when the question requires it. A 2026 H10 study directly compared raw intervals with laboratory ECG in 43 healthy young adults and found excellent agreement in its short resting/reflex protocol. It does not make every recording app, exercise setting or overnight implementation equivalent. Blalock and colleagues.

The first comparison should use simultaneous measurements and matched processing. A morning spot measurement cannot validate an overnight mean from hours earlier. Validation plan.

All measurements · Current targets