A recovery score is a composite, not a measurement
No wearable measures recovery. Recovery is not a physical quantity with units. What these devices measure is a small set of overnight signals — most commonly heart-rate variability, resting heart rate, respiratory rate, and some proxy for sleep duration — and then combine them into a single proprietary number.
The weightings behind that combination are not published, they differ between manufacturers, and they change between firmware versions. Two devices on the same wrist on the same night routinely disagree, and neither is wrong in any checkable sense, because there is no reference standard for a composite that each vendor defines differently.
This matters for interpreting a plateau. A score that will not climb may reflect a genuine physiological change, or a change in how the algorithm weights its inputs, or a shift in your own measurement conditions. The number alone cannot distinguish between those.
What HRV actually reflects
Heart-rate variability is the variation in time between consecutive heartbeats. It is largely a readout of autonomic nervous system activity — the balance between sympathetic and parasympathetic input to the heart.
Because that balance responds to a very long list of inputs, HRV is sensitive but not specific. Alcohol, an argument, a late meal, a developing infection, ambient temperature, hydration, and the position you slept in will all move it. So will training. The signal is real; the attribution is the hard part.
It is also strongly individual. Absolute HRV values vary several-fold between healthy people of the same age and fitness, which is why comparing your number to someone else's tells you close to nothing. Within-person trends over weeks are the only comparison that carries information.
Why measurement conditions explain more plateaus than people expect
HRV is typically sampled during sleep, and what the device captures depends on when in the night it samples, how well it detects sleep onset, and how tightly it is worn. A device that begins sampling earlier in the night will tend to report different values than one sampling in the last hours before waking.
Consistency in the boring variables — same device, same placement, same approximate bedtime, same alcohol and caffeine pattern — removes more noise from a stalled trend than any intervention. Before concluding that a plateau reflects your physiology, it is worth ruling out that it reflects your protocol.
Training load, and the difference between fatigue and maladaptation
A suppressed recovery metric during a hard training block is the expected response, not a malfunction. Adaptation requires a stimulus large enough to disturb homeostasis; the disturbance is visible in exactly these signals.
The published literature on overreaching and overtraining is clearer about the extremes than the middle. Frank overtraining syndrome has recognisable features. The far more common state — accumulated fatigue that has not yet become dysfunction — does not have a validated biomarker, and no consumer score has been shown to identify it reliably.
What this research does not establish
- No consumer recovery score has been validated against a clinical outcome. They are not diagnostic and are not regulated as diagnostic devices.
- Much of the peptide literature relevant to tissue repair is animal work. Every study summary on this site states whether the work was done in animals; a rat tendon model does not establish a human effect.
- Association is not causation. A metric moving alongside a change in routine does not establish that the routine caused it.
Common questions
- Why won't my recovery score improve?
- Most commonly because the score is a proprietary composite of several overnight signals, and any of them can hold it down — including measurement inconsistency rather than physiology. Sustained training load, alcohol, illness, and changes in when the device samples all move the inputs. Because the weighting is unpublished and vendor-specific, a flat score cannot by itself tell you which input is responsible.
- Why am I always tired after working out even though I sleep enough?
- Sleep duration is only one input to how recovered you feel, and it is the one most easily measured, which makes it over-weighted in interpretation. Sleep continuity and timing, cumulative training load, energy availability, and illness all contribute. There is no validated consumer measurement that isolates which of these is dominant for an individual.
- What does low HRV despite good sleep actually mean?
- HRV reflects autonomic balance, which responds to many inputs besides sleep — alcohol, infection, psychological stress, meal timing, ambient temperature, and training. Good sleep does not override those. HRV is a sensitive but non-specific signal, so a low reading indicates that something is loading the system, not what that something is.
- Why am I not bouncing back from training the way I used to?
- Recovery timelines in published studies vary with training status, the nature of the stimulus, age, and energy availability, and the individual variation within any study group is wide — so there is no universal figure to compare yourself against. A change in how quickly you recover can also reflect accumulated training load rather than any single cause, and no consumer metric has been validated to distinguish those. Within-person trends across weeks are more informative than any published average.