Recovery

What does your recovery score actually mean?

It's built from three signals you can name. The number it produces is nearly meaningless on its own — and knowing why is the difference between a score you act on and a score that just makes you anxious.

Every recovery app gives you a number between 0 and 100. Whoop calls it Recovery. Oura calls it Readiness. Garmin has Body Battery. They rarely agree with each other, they sometimes disagree with how you feel, and almost none of them explain what the number is made of.

Here's the part that matters: a recovery score is a weighted summary of three or four physiological signals, compared against your own history. Not against other people. Not against a chart. Against the version of you from the last week or two.

Once you understand that, most of the confusion dissolves — including why your score can read "red" on a morning you feel fine, and why your friend's 90 and your 38 are not comparable numbers.

The signals every recovery score is built from

1. Heart rate variability (HRV)

HRV is the variation in time between consecutive heartbeats. A heart beating at 60bpm isn't ticking once per second like a metronome — the gaps vary, by milliseconds. That variation is largely driven by your parasympathetic nervous system, the "rest and digest" side.

More variation generally indicates a nervous system that isn't under load. Less variation suggests it's dealing with something: training stress, illness, alcohol, poor sleep, a hard week at work. Your body doesn't distinguish between those sources, which is exactly why HRV is useful and exactly why it can't tell you what is wrong.

The thing nobody tells you: HRV isn't one measurement. Apple Watch reports SDNN — the standard deviation of intervals between beats. Most dedicated straps report RMSSD, which looks at the difference between each beat and the next. They are different calculations of the same underlying idea, and 35ms on one is not 35ms on the other. There's no conversion factor. This is the single biggest reason two devices "disagree".

2. Resting heart rate (RHR)

Simpler and more stable than HRV. Your resting heart rate creeps up when your body is working harder than usual at rest — fighting an infection, metabolising alcohol, recovering from an unusually hard session, or carrying accumulated fatigue.

RHR moves in a narrower band than HRV, which makes it less sensitive but more trustworthy. A 5bpm jump above your normal is a genuine signal. A 5ms HRV swing is often just noise.

3. Sleep

Most apps weight sleep heavily, and most people assume that means duration. Duration is roughly half of it. The other half is efficiency — how much of your time in bed you actually spent asleep — and architecture, meaning how much of the night was deep and REM rather than light.

This is why seven hours broken into five fragments scores worse than seven unbroken hours. Your watch knows the difference even when your memory doesn't.

4. Recent training load (sometimes)

Some scores fold in what you did yesterday. Most do it badly for lifters, because the load models were built for endurance athletes — they read heart rate and duration. A heavy set of five at RPE 9 barely registers as "strain" in a cardio-derived model, despite being one of the most systemically demanding things you can do.

Why the number itself is nearly meaningless

Here is the part most articles skip. A recovery score is not an absolute measure of anything. It is a deviation from your own baseline, mapped onto a 0–100 scale.

That has three consequences that matter in practice.

Your baseline ≠ their baseline

Comparing recovery scores with a training partner is comparing two different scales that happen to share a range. One person's 60 might reflect the same physiological state as another's 85.

First: you need history before the score means anything. A score computed on day two has almost nothing to compare against. Most systems need one to two weeks of consistent data before deviations are trustworthy, and some will quietly show you a number anyway.

Second: switching devices restarts the clock. New sensor, new measurement window, new metric, new baseline. Your first month on a new wearable is calibration, not insight.

Third: a good day can't be much better than normal. If sitting exactly at your baseline already scores 100, the scale has nowhere to put the mornings when you're genuinely ahead of it. Well-designed scores reserve headroom at the top, which means an ordinary good day reads in the high 80s or low 90s, not 100. That isn't the app being stingy.

Why two apps give different scores from the same body

DifferenceWhat it changes
SDNN vs RMSSDDifferent maths on the same beats. Not interchangeable.
Measurement windowOvernight average vs spot readings across the day.
SensorOptical wrist vs chest strap vs ring. Different noise profiles.
WeightingOne app may weight sleep at 50%, another at 20%.
Baseline lengthA 7-day baseline reacts faster than a 30-day one.

None of these is "wrong". They're different instruments answering slightly different questions. Pick one, measure it the same way every day, and read the trend. Comparing across devices is the one thing guaranteed to tell you nothing.

How to read your score in week 1, week 4 and week 12

Week 1 — don't act on it. You're building the baseline the score will later be measured against. Train as planned. Wear the device consistently, especially overnight. A score this early is a placeholder.

Week 4 — read deviations, not values. By now "low for you" means something. A single low morning after a hard session is expected physiology, not a warning; it should be heading back toward baseline within a day or two. The reading worth acting on is the one that stays down.

Week 12 — read patterns. You now have enough history to see rhythms: the Monday dip after Sunday's long run, the week your sleep fell apart before a deadline. Be disciplined about what counts as a pattern. Three occurrences is a coincidence with a good story attached. Four or more, in a consistent context, is a pattern.

When to ignore your recovery score

Three situations where the number is actively unhelpful:

And one situation where it's at its most useful: when it disagrees with how you feel slightly. Feeling fine with a real, corroborated drop is the case where an objective signal earns its keep — you'd otherwise have trained straight through it.

The question the score doesn't answer

Here's the gap. You now understand what your recovery score is made of and what the number means. You still don't know what weight to put on the bar.

That translation — from "your HRV is 12% below your seven-day average" to "top set comes down 10%, keep the working sets, swap the second compound" — is the part almost no recovery product does. They hand you a diagnosis and stop. You're left doing the sums yourself, in the gym, under the bar.

There are really only four things you can change on a fatigued day: the load on your top set, the number of working sets, which exercises you pick, and how long you take between them. Most days you only need to move one. Decide which before you walk in, write it down, and you won't renegotiate it mid-session.

Veylan does the translation

It reads HRV, resting heart rate and sleep from your Apple Watch each morning, and rewrites today's session before you train — the weight, the volume, the movements — then explains the change in 80 words, citing the signal that drove it. iPhone + Apple Watch.

Download on the App Store

Veylan is a training tool, not a medical device. Nothing here is medical advice. HRV-guided training is an active research area: the evidence for it comes largely from endurance populations, and effects on performance outcomes have been small and inconsistent. We think the signals are worth reading. We don't think they're magic.