Why Recovery Scores Drop

This article covers what published research says about the physiological signals (HRV, resting heart rate, heart rate recovery) that feed into recovery scores, and what is and isn't established about interpreting a single drop. It does not cover specific illness diagnosis or device-by-device scoring formulas beyond what's publicly documented.

A recovery score dropping doesn't point to one cause. The inputs behind these scores, mainly heart rate variability and resting heart rate, are influenced by training load, illness, alcohol, and emotional state, all of which have been studied separately but not as a single unified explanation for what a wearable displays on any given day. Research on the underlying signals shows they're real and physiologically meaningful, but the composite score sitting on top of them is a synthesis that few companies fully disclose.

The score drops, but the tests come back negative

There's a specific kind of confusion that shows up when a recovery or readiness score tanks and nothing else lines up. No fever, no positive test, sometimes not even a bad night's sleep. The number just goes low, and it stays low in a way that feels different from a normal off day.

That gap between 'the score says something is wrong' and 'nothing external confirms it' is where most of the frustration lives. It's also, as it turns out, a gap the underlying research doesn't fully close.

Training load is the best-documented driver, but it's not the only one

Of everything that can move a recovery score, training load has the deepest research trail. Work on overtraining going back decades describes it as driven by too much high-intensity training combined with too little regeneration time, often stacked with other life stressors, and notes there's still no single diagnostic test for it. Instead, researchers look for stress indicators that fail to return to baseline after a rest period, things like a suppressed immune profile or a depressed psychological state.

A related line of research on soccer players found that banal illnesses (a commonly cited marker of overtraining) could often be predicted when athletes exceeded individual thresholds tied to training strain, meaning load times monotony. That's a specific, testable relationship, but it was observed in structured athlete training data, not in someone's day-to-day life outside a monitored program.

This is part of why a recovery score and the raw HRV number behind it can tell different stories on the same day. The score is a synthesis; the raw signal is one input among several.

Emotional state and substances also move the same signal

Recovery scores lean heavily on heart rate variability, and HRV isn't only responsive to physical training. A systematic review of positive affect and HRV found that resting-state, trait-like positive mood was consistently linked to higher vagally-mediated HRV (the RMSSD and high-frequency measures), though the relationship shifted depending on context, stress phase, and whether the positive affect was momentary or sustained. That means a stretch of low mood, even without any physical strain, sits within the same signal pathway a recovery score is reading.

Alcohol has a more direct, dose-related effect on the same measure. A controlled trial found that two standard alcoholic drinks (red wine or plain ethanol) raised heart rate and reduced total HRV by roughly a third, along with a similar drop in the high-frequency power associated with vagal control, while a single drink had no measurable effect. That's a clean, testable mechanism, but it was measured in a small trial of healthy adults over an eight-minute window, not across a full night or in people with existing health conditions.

None of this collapses into one explanation. It's part of why recovery and readiness scores, even from the same company, can diverge on a given morning depending on which inputs they weight.

What the composite score doesn't show you

The evaluation of 14 composite health scores across major wearable brands found that heart rate variability and resting heart rate are the most common ingredients, but it also found real variation in how transparent companies are about their scoring approach. That's a structural reason a dropped score can feel unexplainable: the researchers doing this evaluation were working from public documentation and white papers, and even they found the underlying methodology often unclear.

This is separate from the question of whether the underlying signals (HRV, resting heart rate, heart rate recovery) are meaningful. They are, in the sense that decades of research connect them to training strain, illness, and autonomic function. What's less settled is how faithfully any single composite number translates those signals into a clean daily readout, and how much of that translation is proprietary rather than published. For a deeper look at that translation problem specifically, whether these composite scores have been independently validated is its own open question.

Common questions