Overview
Heart-rate variability (HRV) is the variation in time between consecutive heartbeats. Resting heart rate (RHR) is how fast your heart beats when you are not doing anything. Together they are the closest thing a consumer wearable has to a read on your autonomic state overnight.
Grio treats them as supporting evidence about recovery, never as a diagnosis, a known cause, or a real-time measurement of stress. The coach is not permitted to convert them into a stress score or to claim it knows how you feel.
How it’s calculated
The personal baseline
Both metrics are graded against a rolling 30-day personal baseline. Population tables are not used: normal HRV spans roughly an order of magnitude across healthy adults, so an absolute threshold would mislabel most people.
That baseline is only usable once it is mature: at least 14 observations for the metric. Before then, Grio shows the readings and withholds the interpretation.
Two different questions, two different windows
Grio runs two comparisons over the same two metrics, because they answer different questions.
| Recovery Shift | Sustained Recovery Trend | |
|---|---|---|
| Question | Is today unusual for me? | Has something stayed different for a while? |
| Window | Today against the rolling 30-day baseline | The last 7 local dates against the preceding, non-overlapping 60 |
| Maturity | 14 observations for the metric | 5 valid recent observations and 30 in the reference window |
| Purpose | A same-day observation | Evidence a shifting baseline would otherwise absorb |
Recovery Shift requires two gates at once. An HRV concern must be both at least two standard deviations below baseline and at least 12 percent below it. A resting-heart-rate concern must be both at least two standard deviations above baseline and at least 8 percent above it. Requiring both avoids overreacting to a naturally narrow baseline, and avoids treating a large percentage change built on weak personal evidence as if it were solid.
Sustained Recovery Trend compares averages rather than a single day. Its reference band is the reference-window mean plus or minus the greater of the population standard deviation and three percent of that mean. HRV is concerning only when the recent average is below the band, at least 12 percent below the reference average, and at least four recent observations fall below the band. Resting heart rate mirrors that with an 8 percent threshold in the upward direction. Either metric can establish a trend; both together are corroboration, not a separate severity tier.
While the long comparison is still filling up, Grio shows a Recovery Context Stage rather than an empty state: one to four observed days is Starting Context, five to six is an Early Pattern, seven or more is Recent Context. These stages report what has been observed. They do not assign a normal/abnormal direction and they do not trigger a message.
Freshness and edge cases
- Freshness is part of eligibility, not copy decoration: an observation whose recency cannot be established is kept as history and never presented as a current change.
- Missing metadata never makes an observation fresh:
- A device reporting only one metric carries the whole cardiovascular axis in the Condition Score.
- A single overnight HRV sample is discounted relative to a night sampled repeatedly.
How to read your score
The direction matters more than the number:
- HRV lower than your usual, RHR higher than your usual is the classic loaded pattern. It is consistent with training load, poor sleep, alcohol, illness, and ordinary life stress, all at once. It does not identify which.
- Both inside your usual range means today looks like your other days. It is not a certificate of health.
- HRV unusually high is not automatically good. A very high reading with an elevated resting heart rate is a known pattern worth watching rather than celebrating, which is why Grio treats an isolated favourable reading as bounded rather than decisive.
The single most common misreading is treating night-to-night movement as signal. These metrics are noisy at the daily scale. That noise is exactly why Grio uses two standard deviations and a percentage floor before it says anything.
What to do with it
- When a shift is flagged and you know why (a late night, a hard session, or a glass of wine), the useful action is the obvious one, and Grio will not pretend to have discovered a cause it cannot see.
- When a shift is flagged and you feel fine, your report leads. Grio may add one small recovery-oriented action; it will not require a Check-In or a question before you get value from the observation.
- When the sustained trend flags a change, that is the one worth a second look, because it survived a moving baseline for a week.
Detecting either signal never automatically mutates a committed Daily Plan or any goal field, including weekly pace. It can inform ordinary coaching and a plan you have not committed to yet. Changing something durable stays your call.
Troubleshooting & FAQs
- My HRV dropped 20% but Grio said nothing. Why?
- A percentage change alone is not enough. Grio requires the reading to be both a large statistical departure from your own baseline and a large percentage change, and it requires a mature baseline of at least 14 observations for that metric. Requiring both gates stops it from overreacting to a naturally narrow baseline or to a big percentage swing built on thin evidence.
- Grio noticed a change days after the night it happened. Is that a bug?
- No, that is the freshness rule working. A delayed upload can update your history but must not be presented as a current change. If Grio cannot establish that an observation is fresh, it keeps the data for context and suppresses the message.
- My watch records resting heart rate but not HRV. Is my score worse?
- No. Whichever metric your device reports carries the whole cardiovascular recovery axis. Grio does not redistribute the missing half to sleep or self-report, because that would let the only physiological evidence of the day be outvoted.