A recovery score is a single daily number that answers one question: is my body ready for a hard day? It is calculated from the signals your body already produces overnight — sleep duration and quality, heart-rate variability (HRV), resting heart rate — combined with your recent activity, and compared against your own recent baseline rather than population averages. Higher generally means more recovered; lower means your body is still paying for yesterday.
The signals behind the number
| Signal | What it reflects | How it moves |
|---|---|---|
| Sleep | Duration and structure of last night's sleep — the raw material of recovery | Short or fragmented sleep pulls the score down the next morning |
| HRV (heart-rate variability) | The variation in time between heartbeats — a window on autonomic nervous-system state | Higher than your baseline tends to accompany good recovery; suppressed HRV often accompanies fatigue, illness, alcohol or stress |
| Resting heart rate | How hard your heart works at rest | An elevated resting heart rate versus your baseline is one of the most consistent signs the body is under load |
| Recent activity | Training load over the past days | Hard recent sessions legitimately lower a recovery score — that is the point, not a bug |
Why “your baseline” is the whole game
A recovery score is not a grade. It is a comparison. A resting heart rate of 58 and an HRV of 70 ms might be excellent for one person and unusual for another, so every serious app builds a rolling personal baseline (commonly 7–60 days depending on the signal) and scores you against yourself.
This is also why one low score means very little. A single late night, a hot room, dehydration or an oncoming illness can dip a score for a day. The useful signal is the trend: a score sliding over several days, or a resting heart rate creeping up over weeks, is what deserves attention — and, if persistent, a conversation with a clinician.
How the math typically works
- 1Read the overnight signals. Each morning, the app reads sleep, HRV and resting heart rate from Apple Health.
- 2Compare with your baseline. Each signal is expressed as a deviation from your recent personal average.
- 3Weight and combine. The deviations are weighted into a single score, often shown as a category (for example Good / Stable / Strained) or a 0–100 scale.
- 4Account for recent load. Recent activity is factored in so a hard training block reads as intentional load, not a broken body.
Different apps make different choices at step 3 — which HRV statistic to use (RMSSD or SDNN), how long a baseline to build, how much weight to give resting heart rate versus sleep. That is why the same morning can produce a 70 in one app and a 55 in another. Scores are only comparable within one app. Pick one and follow its trend.
What a recovery score is not
- Not a diagnosis. Consumer wearables estimate, they do not measure clinically. A recovery score cannot tell you what is wrong.
- Not a leaderboard. Comparing your score to other people's is meaningless — baselines differ by orders of magnitude between individuals.
- Not a command. A low score is information, not an instruction. Life happens; the score just describes the trade.
When the number misleads
A recovery score summarizes last night — and last night is not always a fair test. Several well-documented situations move the underlying signals without any change in your actual fitness:
- Alcohol. Even when you sleep a full eight hours, alcohol suppresses HRV and can elevate resting heart rate through the night. A score the morning after reflects the drinking, not your training load.
- An oncoming illness. Resting heart rate often rises and HRV falls in the day or two before you feel symptoms. A cluster of low scores with no training explanation is one of the more useful early warnings a wearable can give.
- Travel and time zones. A red-eye or a shifted sleep window changes both sleep structure and the overnight heart-rate pattern. Jet-lag weeks produce low scores that say nothing about fitness.
- Menstrual cycle. HRV measurably shifts across the cycle in people who menstruate, with a dip around menstruation in many studies. Baselines that do not account for this can read normal variation as poor recovery.
None of this makes the score useless — it makes a single day's score weak evidence. The pattern over several days, read alongside how you actually feel, is the signal worth acting on. If low scores persist for a week or more without a clear cause, or come with symptoms, that is a conversation for a clinician.
Where the underlying signals come from
The physiology behind these signals is well studied in the clinical literature: resting heart rate tracked over years is associated with long-term mortality risk in cohort studies (for example, Jensen et al., Heart, 2013, which followed 2,798 men for 16 years and found each 10 bpm faster resting heart rate associated with about 16% higher mortality risk — an association that persisted after adjusting for physical fitness). HRV is widely used in research as a non-invasive window on autonomic function. These are associations from observational data — useful context, not personal medical conclusions.
The evidence behind the inputs
Each individual signal feeding a recovery score has decades of research behind it. The composite score does not — no study validates any particular app's weighted formula, and honest apps say so. What the literature actually supports:
- Resting heart rate. Jensen et al. (Heart, 2013): each 10 bpm faster resting heart rate was associated with about 16% higher mortality risk over 16 years in 2,798 men — an association that persisted after adjusting for physical fitness.
- HRV. A 2022 meta-analysis (Jarczok et al.) found people in the lowest quartile of HRV had a 56% higher hazard of all-cause mortality (HR 1.56, 95% CI 1.32–1.85) than the higher quartiles. See what a normal HRV range looks like.
- Sleep regularity. Two UK Biobank accelerometer studies totalling roughly 150,000 people found more consistent sleep timing tracked with lower all-cause mortality — a larger and more consistent effect than sleeping longer. See the sleep-regularity research.
All of these are observational associations — they show that the signals correlate with outcomes, not that moving one input changes your personal risk. The recovery score is best understood as a convenient summary of well-studied signals, read as a trend against your own baseline. If scores stay low for a week or more without a clear cause, or come with symptoms, that is a conversation for a clinician.