Smartwatch Recovery Scores: How They're Calculated and Why They're Not Exact

Short answer: the recovery or readiness score on your watch, ring or band is an estimate, not a measurement. It blends a few inputs (overnight heart rate variability, resting heart rate, sleep and recent training). The device measures each with some error, and the formula that combines them isn't published. Researchers have only validated a small share of the devices on sale for even one measure (1), and since there is no single lab test for "recovery", there is nothing direct to check the score itself against. Used as a trend against your own normal, it can help. Used as a verdict on a single morning, it can mislead. So look at the trend and, as a rule of thumb, change a hard session only after two low mornings in a row, or when a low morning matches how you feel.

This matters because many athletes now let a coloured ring decide whether today is hard or easy. That is a lot of weight for a number with a margin of error you can't see.

How should you use your recovery score?

The recommendation for most amateurs: treat the score as one voice in the room, not the decision. Look at the trend over several days, against your own baseline, and check it against how you feel and how your last sessions went.

In practice:

• Look at the inputs, not just the score. Most apps let you see overnight HRV (heart rate variability: the tiny beat-to-beat changes in your heart rhythm, which tend to shrink when you are stressed, ill or tired), resting heart rate and sleep separately. A low score caused by one short night means something different from a low score caused by a rising resting heart rate.
• Know your normal range. It is your average over the past few weeks, plus or minus your usual day-to-day bounce. Many apps shade it for you. For example, if your HRV usually sits between 55 and 70, a reading of 48 is low. If your app shows no range, note your readings for two or three weeks and work it out from those.
• Act on patterns, not one reading alone. As a rule of thumb, change a hard session when the score or its inputs are below your range for two mornings in a row, or when one low morning matches how you feel (heavy legs, poor sleep, a cold coming on). The change can be swapping intervals for an easy session or cutting them short. This is a coaching rule of thumb, not a study finding.
• Trust the device more for some things than others. Heart rate is reasonably good, and sleep duration is usable if you remember devices tend to overstate it. Sleep stages, VO2max estimates and calorie counts are rough (1, 2).
• Keep the comparison fair. Same device, worn the same way, every night. A new device, a new strap or a software update can shift your numbers, so give it a couple of weeks before reading too much into changes.

What matters most depends on where you are:

• If you're new to structured training, the score is most useful as a nudge to notice sleep, alcohol, illness and stress. Don't let it talk you out of easy sessions when you feel well. The exception is illness: with a fever or chest symptoms, rest whatever the score says.
• If you're experienced, use the inputs to time your hardest sessions, and watch for low readings that keep repeating across a hard block.

What goes into a recovery score?

Most scores combine some of these:

• Overnight HRV (heart rate variability): the small beat-to-beat changes in your heart rhythm, which tend to fall when you are stressed, ill or tired.
• Resting heart rate, usually the lowest or average overnight value.
• Sleep: how long you slept, sometimes how much time in each sleep stage.
• Recent training load, and sometimes skin temperature, breathing rate or how you rated your day.

The device scores each input against your own recent history and blends them into one number. The weighting is proprietary. A review of wearables concluded that their algorithms are often not publicly available or lack high-quality validation, and advised caution in interpreting their metrics (3). So when two devices disagree about the same night, they may both be "working as designed".

How accurate is each input?

Heart rate: good. Pooled across many validation studies and conditions, wearables measured heart rate with an average bias of about 3% (1). A small study of one wrist band against a clinical ECG (the hospital-grade heart recording) during sleep found its heart rate agreed well (4).

HRV: decent, but the error can be close to the size of the changes you care about. In short: methods that have been checked against ECG are generally usable, and unchecked ones can be badly wrong. The detail: a meta-analysis of short recordings, mostly at rest, found HRV from portable devices differed from ECG by a small amount, with large variation between studies (5). In one study, a chest strap and a validated smartphone camera app both agreed closely with ECG at rest (6). But methods differ: in a small test of seven products on five people, the HRV error was about 4% for the best (an app reading a chest strap), about 7% for the one ring tested, and over 100% for another app using the phone camera (7). The takeaway: use a method that has been checked against ECG. For the wrist band study above, the HRV error approached or exceeded the smallest change worth paying attention to (4).

That matters because the effects you are looking for are small. In a large analysis of millions of morning readings, after hard training days HRV fell by only about 5% and resting heart rate rose by about 1% on average; heavy drinking and illness had bigger effects, around a 10–12% drop in HRV and a 6% rise in resting heart rate (8). HRV also bounces from morning to morning on its own, even when measured perfectly (4). When the signal from training is a few per cent, natural bounce plus a few per cent of device error can hide it or invent it. Hence the rule: look at trends over days, and don't act on one reading alone.

Sleep duration: reasonable. Sleep stages: rough. In a lab study of six devices (worn on the wrist, finger or forehead) against polysomnography (the full overnight sleep test), all six were valid for when you slept and for how long. Agreement on sleep stages was only 50–65% (2). Another lab study of seven devices, some worn and some placed by the bed, found they were very good at detecting sleep but much less good at detecting when you were awake, and worse on disrupted nights (9). Reviews found devices tend to overestimate total sleep time, with errors typically above 10% (1); one meta-analysis found older models overestimated it, while newer ones did better (10). So a "light sleep" or "deep sleep" figure is a rough estimate, and a bad night may look better than it was.

VO2max: an estimate with wide error. VO2max is the most oxygen your body can use. It isn't usually part of the recovery score, but it is another estimate many watches show. A meta-analysis found that devices estimating it from resting data overestimated it, while those using exercise data (heart rate against pace or power) did better on average. Even then, the individual error could be around 10 mL/kg/min either way (11). For an athlete with a VO2max of 50, that is a fifth of the number. The authors concluded the estimate works for groups but is too imprecise for individuals for sport purposes (11).

How to use this in a real training week

Say you have intervals on Tuesday, a long ride on Saturday and easy sessions in between.

Tuesday morning, the score is low. Open the inputs. If it is low because you slept five hours after a late night, and resting heart rate and HRV are in your normal range, you can usually train, but consider doing the intervals at the lower end of your targets. If HRV is down and resting heart rate is up by a few beats, and you feel flat, swap the intervals for an easy session and move them to Wednesday or Thursday.

The score is high, but your legs are heavy. Believe your legs. A green score is not an instruction to add extra work.

The score has been low for a week. First check whether your training load has risen. As a rule of thumb, cut back to easy sessions for three to five days; if your numbers are back in range, build up again. If they aren't, look outside training: sleep, work stress, alcohol, illness, and whether you are eating enough for your training (a sports dietitian can help with that). If you feel unwell, or notice chest pain, palpitations or unusual breathlessness, see a doctor rather than relying on a wearable.

You just changed device or updated the software. Your baseline may shift. A review of sleep wearables notes that companies can change their algorithms without notice, and lists firmware updates among the factors to watch (12). Give it a couple of weeks before comparing with old numbers.

Tito uses the same inputs, HRV, resting heart rate and sleep, and shows each one in real units against your own range, so you can see which input moved rather than just a single score.

A simple routine for trustworthy numbers

This is the recommended version for most amateurs:

1. Wear the device the same way every night, snug but comfortable; fit affects optical readings (3).
2. Collect at least two to three weeks of readings before trusting your baseline.
3. Each morning, check three things: overnight HRV against your range, resting heart rate against your range, and sleep duration. Ignore sleep stages.
4. Change the plan only when inputs are below your range for two mornings, or when one low morning agrees with how you feel.
5. Write one line in your training log: how you slept, how you feel, anything unusual (late meal, alcohol, travel). Over a few weeks this tells you what really moves your numbers.

Other options

Measure HRV yourself in the morning. A short reading on waking, a few minutes lying or sitting still in the same position each day, with a chest strap or a validated phone app, gives a standardised measurement that agreed closely with ECG in testing (6).

Use a chest strap for training heart rate. Many coaches prefer a chest strap for intervals, where wrist optical sensors are more prone to error (3). If your wrist readings jump around during hard efforts, a strap is the simple fix.

Ignore the score entirely. If you find yourself anxious about the number, or it keeps contradicting how you feel, it is reasonable to stop looking at it and train on perceived effort and a simple log.

The science behind it

This part is optional reading for anyone who wants the evidence.

The scale of validation is the first problem. An umbrella review pooling 24 systematic reviews and 249 validation studies estimated that only about 11% of consumer wearables on the market had been validated for at least one measure, and that the studies done so far cover only about 3.5% of what a full evaluation would need (1). The same review found heart rate was off by about 3% on average, VO2max estimates had wide errors, larger from resting tests than from exercise tests, and total sleep time was usually overestimated (1).

For sleep, lab comparisons with polysomnography find that consumer devices detect sleep well and wake poorly, and reviews find they tend to overstate how long you slept (1), though in these lab studies only some devices did (9, 2). A meta-analysis of one brand found that newer models using heart rate and movement did better than older ones using movement alone, but still concluded they give "gross estimates" and are not a substitute for a sleep lab (10).

For HRV, the meta-analysis of portable devices found small but variable errors against ECG, with the body position during measurement and the HRV metric used both affecting the error (5). Small differences matter here because the day-to-day change after a hard training day is itself small, about 5% on average, though alcohol and illness move HRV more (8). That is the core reason to read HRV as a weekly trend.

We found no published study validating a recovery score as a measure of "recovery" itself, and there is no single lab test for it to be checked against. A score can only be as good as its inputs and the formula that combines them, and wearable algorithms are often not public (3).

References

1. Doherty C, Baldwin M, Keogh A, et al. Keeping Pace with Wearables: A Living Umbrella Review of Systematic Reviews Evaluating the Accuracy of Consumer Wearable Technologies in Health Measurement. Sports Med. 2024. doi:10.1007/s40279-024-02077-2
2. Miller DJ, Sargent C, Roach GD. A Validation of Six Wearable Devices for Estimating Sleep, Heart Rate and Heart Rate Variability in Healthy Adults. Sensors (Basel). 2022. doi:10.3390/s22166317
3. Shei RJ, Holder IG, Oumsang AS, et al. Wearable activity trackers-advanced technology or advanced marketing?. Eur J Appl Physiol. 2022. doi:10.1007/s00421-022-04951-1
4. Bellenger CR, Miller DJ, Halson SL, et al. Wrist-Based Photoplethysmography Assessment of Heart Rate and Heart Rate Variability: Validation of WHOOP. Sensors (Basel). 2021. doi:10.3390/s21103571
5. Dobbs WC, Fedewa MV, MacDonald HV, et al. The Accuracy of Acquiring Heart Rate Variability from Portable Devices: A Systematic Review and Meta-Analysis. Sports Med. 2019. doi:10.1007/s40279-019-01061-5
6. Plews DJ, Scott B, Altini M, et al. Comparison of Heart-Rate-Variability Recording With Smartphone Photoplethysmography, Polar H7 Chest Strap, and Electrocardiography. Int J Sports Physiol Perform. 2017. doi:10.1123/ijspp.2016-0668
7. Stone JD, Ulman HK, Tran K, et al. Assessing the Accuracy of Popular Commercial Technologies That Measure Resting Heart Rate and Heart Rate Variability. Front Sports Act Living. 2021. doi:10.3389/fspor.2021.585870
8. Altini M, Plews D. What Is behind Changes in Resting Heart Rate and Heart Rate Variability? A Large-Scale Analysis of Longitudinal Measurements Acquired in Free-Living. Sensors (Basel). 2021. doi:10.3390/s21237932
9. Chinoy ED, Cuellar JA, Huwa KE, et al. Performance of seven consumer sleep-tracking devices compared with polysomnography. Sleep. 2021. doi:10.1093/sleep/zsaa291
10. Haghayegh S, Khoshnevis S, Smolensky MH, et al. Accuracy of Wristband Fitbit Models in Assessing Sleep: Systematic Review and Meta-Analysis. J Med Internet Res. 2019. doi:10.2196/16273
11. Molina-Garcia P, Notbohm HL, Schumann M, et al. Validity of Estimating the Maximal Oxygen Consumption by Consumer Wearables: A Systematic Review with Meta-analysis and Expert Statement of the INTERLIVE Network. Sports Med. 2022. doi:10.1007/s40279-021-01639-y
12. de Zambotti M, Cellini N, Goldstone A, et al. Wearable Sleep Technology in Clinical and Research Settings. Med Sci Sports Exerc. 2019. doi:10.1249/mss.0000000000001947