Garmin Metrics
Body Battery, HRV Status, Sleep Score: Three Metrics or One?
August 1, 2026
Are Body Battery, HRV status and sleep score three different metrics?
Functionally, no. In our data they behave like one underlying autonomic signal presented through three interfaces, and the practical consequence is that when all three agree on a Monday morning, you have not received three-way confirmation. You have received one measurement, restated three times.
That distinction changes how much weight the morning dashboard deserves. Most athletes I talk to treat agreement between Body Battery, HRV status and sleep score as corroboration. It is closer to reading the same sentence in three fonts.
What we measured
Our platform stores daily Garmin values for athletes who connect their accounts. For study S4 (computed 2026-08-01) I paired each athlete's same-day values, computed a Spearman rank correlation within each athlete, then looked at how those correlations were distributed across the cohort. Each athlete needed at least 30 paired days. Six accounts were excluded as contaminated: created inside the 2026-05-03 data-leak incident window with near-identical multi-year histories under distinct Garmin accounts.
Two definitions matter. body_battery_high is the day's peak, which for almost everyone lands in the morning after the overnight recharge. hrv is the overnight average that feeds that same morning. This is a comparison of two readouts of the same night.
Within-athlete is the right way to ask the question, because comparing raw HRV across people mostly measures the fact that some people run higher HRV than others. It also limits what I can claim: everything below is about how your numbers move against your own baseline.
The numbers
| Pair | Median rho | p25 to p75 | Athletes | Share past ±0.3 |
|---|---|---|---|---|
| Body Battery peak vs overnight HRV | 0.674 | 0.574 to 0.793 | 82 | 97.6% above 0.3 |
| Body Battery peak vs sleep score | 0.683 | 0.618 to 0.760 | 85 | 98.8% above 0.3 |
| HRV vs resting heart rate (control) | -0.696 | -0.767 to -0.583 | 82 | 95.1% below -0.3 |
| Body Battery vs stress average (control) | -0.452 | -0.538 to -0.370 | 93 | 81.7% below -0.3 |
The striking line in that table is not a median. It is that 100% of the 82 athletes showed a positive Body Battery-to-HRV correlation. Not one person had a watch where higher overnight HRV went with a lower morning Body Battery. Across 82 humans with different sleep habits, watch models and training loads, unanimity is rare. It is what you see when two numbers share a source.
The controls confirm the method is measuring something real rather than manufacturing agreement. HRV against resting heart rate came out at -0.696, which is textbook autonomic physiology: nights where the parasympathetic system is dominant show higher variability and a slower resting pulse. If our pipeline were generating spurious positive correlations, that pair would not have landed cleanly negative in 95.1% of athletes.
The stress control is the interesting one. Body Battery is built from stress by construction, so a negative relationship is expected, and -0.452 delivers it. But notice that the by-construction input correlates less strongly with Body Battery than HRV and sleep score do. My reading, which the data supports but does not prove: the day's peak is set almost entirely overnight, so overnight inputs dominate it, while daytime stress governs the shape of the drain curve after the peak. If you want the mechanics of that drain, our guide to what Garmin's stress score actually measures covers it.
Why this is one signal, not three opinions
Correlations of 0.67 within an individual are high for physiology, and the mechanism behind them is not a secret. Garmin documents that Body Battery is computed from heart rate variability, stress, sleep and activity. Sleep score is built from sleep duration, sleep stages, awake time and overnight stress, and Garmin's stress figure is itself derived from HRV. HRV status compares your overnight HRV against your own baseline range. Three definitions, one primary ingredient, measured over the same eight hours. It would be strange if they disagreed.
So the three tiles on your morning screen are better understood as three renderings of one overnight autonomic measurement: one as a fuel gauge, one as a status word, one as a score out of 100. Different units, same underlying reading.
The stacking mistake
This is where it costs you something. Say you wake up to Body Battery 34, HRV status "unbalanced" and sleep score 51. The instinctive read is that three independent systems have converged on the same conclusion, which feels like strong evidence, so you cancel the workout with high confidence.
It is not strong evidence. It is one measurement of one night, rendered three ways, and it carries exactly the confidence of a single overnight HRV reading. Overnight HRV is useful and also noisy: a late meal, a warm room, a glass of wine, or a watch that slid down your wrist can move it. Three displays of a noisy measurement do not average out the noise. They multiply the appearance of certainty while adding nothing.
Stacking agreeing metrics as though they were independent evidence is the most common way I see people over-read their morning screen. The fix is arithmetic, not willpower: count the trio as one input, not three.
Our own complete Body Battery guide tells you to check the number before deciding workout intensity, and our troubleshooting guide tells you to use Training Readiness for training decisions and Body Battery for life. I still think both are right. What neither said, and should have, is how much of the same information those metrics are carrying. That is a gap in our earlier writing, and this article is the correction.
This is not a scandal
"These metrics are all the same" is the kind of line that reads as an exposé, and it should not. Garmin never claimed independence, and the documentation describes the inputs plainly. A coherent recovery model is arguably a good sign: if your watch told you that your HRV was excellent, your Body Battery was drained and your sleep was terrible on the same morning, the sensible conclusion would be that at least one of them is broken. Internal consistency is what you want from a measurement system.
My criticism is narrower and it is about presentation. Three separately named, separately branded, equally weighted tiles on one screen invite a confirmation reading that the underlying math does not support. A dashboard that showed one recovery reading with three views behind it would be describing the same physiology more honestly than one that shows three cards in a row.
Coherent is not the same as predictive
One thing agreement definitely does not buy you is accuracy about today. In a separate study on the same platform, we correlated the morning readiness score built on these signals against same-day running performance across 5,136 steady runs from 39 athletes. Median within-athlete correlation: 0.056. Only 13% of athletes cleared 0.3. Splitting runs into readiness quintiles moved median performance by about 0.15 of a standard deviation from worst to best, which is close to nothing.
So the trio is a consistent description of your night and a poor forecast of your next run. Confusing those two properties is how people end up skipping sessions they would have executed fine. If you have ever wondered why your Training Readiness runs low on days you feel strong, that null is the reason. We hit the same pattern of "measuring something, just not the thing you assumed" when we tested Garmin's race predictor against real races.
What to do with your morning screen instead
Count the trio once. When Body Battery, HRV status and sleep score agree, log that as one recovery data point of moderate confidence. Adjust your session at the margins, not by cancelling it outright.
Treat disagreement as the informative case. Because these numbers normally track together at 0.67, a morning where they split is unusual and worth two minutes of attention. Usually it means a data-quality problem: sensor contact lost, a partial night recorded, the watch on the charger for part of the night. Our Body Battery troubleshooting guide covers the common causes.
Get a second opinion from a different axis. How a warmup feels at a fixed pace, morning body weight, whether yesterday's session left soreness that is still there, what you know about your own life stress. None of those come from the overnight sensor, which is what makes them worth having.
Watch the direction over a week. A five-day slide in your morning peak is a real signal; Tuesday's number against Monday's is mostly noise. Our guide to reading sleep data as an athlete walks through what to track over that window.
Caveats you should hold this with
Correlation is not causation, and nothing here identifies a direction. These are same-day pairings of an overnight window with the morning peak that follows it, so they say nothing about lag effects across days. The cohort is self-selected: Garmin users who chose to connect a training-analytics platform, likely fitter, more data-curious and better at wearing the watch than the general Garmin population. And every number is a within-athlete correlation. That is the right question for a personal dashboard, and it means none of this supports claims about how athletes compare to each other.
Frequently Asked Questions
Is Body Battery just HRV with extra steps?
Not quite, but closer than the interface suggests. Body Battery adds activity and daytime stress, which is why it drains through the day while HRV status sits still. For the morning peak specifically, our data shows a median within-athlete correlation of 0.674 with overnight HRV, positive in all 82 athletes measured. The extra steps mostly shape what happens after the peak.
Should I turn two of the three off?
Only if the clutter bothers you. There is no accuracy penalty for viewing all three. The penalty is psychological: three tiles agreeing feels like more evidence than one number, and it is not.
If they measure the same thing, why do mine sometimes disagree?
That is worth investigating. In our cohort, 97.6% of athletes had a Body Battery-to-HRV correlation above 0.3, so disagreement is the exception rather than the rule. The most common causes are data-quality problems overnight: the watch losing skin contact, being off the wrist for part of the night, or a sleep window the algorithm failed to detect properly.
Does this mean Body Battery is inaccurate?
That is a different question from the one this article asks. Nothing here tests whether Body Battery correctly describes your energy; it tests whether it tells you something your other recovery metrics do not, and mostly it does not. Separately, we found that a readiness score built on these signals had a median correlation of 0.056 with same-day running performance, which is the accuracy problem worth worrying about.
What is actually independent of the overnight signal?
Anything not derived from your wrist between midnight and 6am. Perceived effort at a known pace, body weight and hydration, whether soreness from a hard session is still there on day three, and the non-physiological stuff like a deadline week. Those add information rather than restating it.
What would change my mind
The finding that would break this article is a firmware or algorithm change that makes the three numbers diverge, with Body Battery picking up something HRV and sleep score cannot see. If Garmin adds an independent input with real weight, the within-athlete correlations should fall and the unanimity across 82 athletes should break. That is a measurable prediction, and we will re-run S4 against the same cohort when a major Firstbeat revision ships.
Until then, the honest instruction for your morning screen is short. One night, one reading, three fonts. Weight it accordingly, and go find your second opinion somewhere your watch cannot reach.
