What moves your HRV: the six levers that matter for HYROX athletes
Your variability score does not fluctuate randomly. Six specific inputs drive it in consistent, measurable ways. Understanding which one is doing what changes how you interpret the number every morning.
HYROX Rome 2025. I arrived sick, I raced sick. The evening after the finish, I had a glass of white wine. One glass.
The Garmin reading the following morning: sleep score 37, Body Battery 4, training readiness 14. My variability score was 51. My baseline is 93.
That morning is the clearest illustration I have of how levers stack. The illness was one input. The race effort was another. The one glass of wine was a third. Each one alone would have produced a low reading. Together, they produced the floor. The number I saw on my watch was not one thing going wrong. It was three things going wrong simultaneously, and the watch gave me a single number that told me nothing about which one to fix first.
If you have read what HRV actually measures, you already know what the number is tracking. A variability score, or RMSSD, reflects the state of the tug-of-war between your activation system (the sympathetic nervous system) and your recovery system (the parasympathetic nervous system). High variability means both systems are engaged and balanced. Low variability means the recovery system has stepped back.
What the number does not tell you is which specific input is pulling the rope today. And that matters. A low variability score caused by two drinks at dinner looks identical on screen to one caused by three consecutive hard sessions. The number is the same. The appropriate response is completely different.
These are the six inputs that reliably move the number, in rough order of magnitude.
Alcohol: the most consistent and predictable lever
I almost never drink. One glass of wine on a race evening in Rome is about as close as I get to regular alcohol consumption. Which makes the effect easier to see when it happens: the morning after, no energy, a heart that feels heavy and struggles to pump efficiently, legs that are half asleep. And a variability score that went from a race-week low to a reading I had never seen before.
The Rome morning was a compound scenario: illness, race effort, and alcohol stacked in 24 hours. But the alcohol was the lever that took a predictably low reading and made it catastrophic. Without it, a sick athlete the morning after a HYROX race would read low. With it, I was at 51 against a baseline of 93.
What the science says
The claim that moderate drinking does not affect recovery does not hold in the data. Pietilä et al. (2018) tracked overnight variability scores across 4,098 Finnish employees across thousands of nights [1]. Even moderate consumption, one to two standard drinks, produced a statistically significant reduction in overnight variability. The effect was dose-dependent: more drinks, larger drop, longer suppression. More importantly, alcohol shifted the autonomic nervous system toward sympathetic dominance during the part of the night when your recovery system should be running unchallenged.
The duration is what catches athletes. The suppression does not resolve by morning. In most subjects, variability remained below personal baseline for 12 to 24 hours after moderate consumption. For heavier consumption, that window extended to 48 hours. A Friday evening with friends produces a suppressed Saturday reading and often a partially suppressed Sunday, covering an entire training weekend before you have done a single session.
If you drink, expect a low variability score the following night, and possibly the one after. Do not use that reading to adjust your training plan. The number is telling you about the alcohol, not about your readiness. Track the two variables separately.
Sleep: not just hours, but what happens inside them
One night of 6 hours 30 minutes against my average of 7 hours 30. The next morning, my orthostatic variability score was 67 against a baseline of 93. A drop of 26 points from a single short night, with no training load, no alcohol, no illness as a confounding variable. One lever, clean signal.
Sleep is when your variability score is measured. The nocturnal reading captures exactly this: how well your recovery system dominated while you were not interfering with it. If the sleep was cut short or fragmented, the reading reflects that directly, not your training readiness, not your fitness. The sleep.
What the science says
Fullagar et al. (2015) reviewed the effects of sleep restriction on athletic performance across multiple controlled studies and confirmed that sleep loss consistently impairs the physiological and cognitive functions most relevant to training quality [2]. The variability signal follows the same pattern: less sleep, lower score the following morning.
What is less often discussed is sleep architecture. The quality of slow-wave sleep, the deepest non-REM stage where tissue repair and hormonal recovery run, matters independently of total hours. A 7-hour night with multiple wake periods can produce a lower variability score than a clean 6-hour night, because slow-wave sleep was interrupted and the recovery system never ran at full capacity. Your wearable measures time in bed reliably. What disrupts architecture (alcohol, late meals, ambient noise) it captures only indirectly through the resulting variability signal.
The rule that counts
When your variability score is low, check your sleep score before interpreting the reading as a training signal. A short or fragmented night is its own explanation. The priority becomes the next night, not the current session.
Acute training load: the drop is not the signal
I remember training blocks where a day’s three sessions were planned without accounting for what had already accumulated across the week. The load made sense on paper. In practice, the cumulative weight of what had already been done made those three sessions a different kind of effort altogether. The next morning, I was flat. Not sore, not tired in a manageable way. Flat. A full rest day that had not been planned became necessary.
That kind of day illustrates the distinction that matters with training load and HRV: the drop after a hard session is expected and appropriate. Your activation system has been dominant. Your recovery system needs time to restore its share. The useful signal is not the drop. It is how long recovery takes.
What the science says
Treating each day’s reading in isolation does not work as a decision framework. The drop itself is expected physiology, not a signal to act on. What does not resolve as expected is what matters. Plews et al. (2013) established the reference window in elite endurance athletes: variability returns to personal baseline within 48 to 72 hours after a high-intensity effort in a well-adapted athlete under normal load [3]. A reading that stays significantly below baseline beyond that window is where the information starts.
Buchheit’s 2014 review confirmed that the weekly trend is the meaningful unit of analysis, not the daily number [4]. A variability score declining progressively across consecutive training blocks without a scheduled rest week is the overreaching signal. A single hard session producing a low reading the next morning is just the training working.
A low variability score 12 to 24 hours after an intense session is not information worth acting on. Still below your normal range at 72 hours, especially if the pattern repeats across multiple blocks, is worth attention. Track the weekly trend, not the daily snapshot.
Non-training stress: your nervous system does not categorise load by source
Your nervous system does not distinguish between the stress of a hard interval session and the stress of a difficult work week. Both activate the same system. The variability score reflects total load, not just the training component. For athletes training alongside a full-time professional life, that distinction is load-bearing.
The mismatch shows in both directions. A light training week where the score stays suppressed usually has a non-training explanation: a heavy work period, disrupted travel, the early stages of an infection, or accumulated sleep debt. A hard training block where the score holds steady is often a week where everything else was well managed. The number is accounting for all of it.
What the science says
Thayer et al.’s meta-analysis of 31 neuroimaging studies established that the autonomic nervous system, measured via HRV, is directly coupled to brain regions involved in emotional regulation and threat processing [5]. Psychosocial stress, sustained cognitive load, and emotional arousal all measurably reduce parasympathetic activity through the same pathways as physical training stress.
The practical implication is direct. A genuinely heavy work period, sustained travel, or a significant personal stressor can suppress your variability score as reliably as a loaded training block. The appropriate response is completely different. Adjusting your training load in response to a work-stress HRV dip is usually the wrong move. Protecting sleep and recovery inputs becomes the primary lever. The session is not the problem.
The rule that counts
When your variability score is low and your training log does not explain it, look at the full picture: work load, travel, sleep quality, early signs of illness. The number reflects total nervous system load. Adjust accordingly, not automatically toward less training.
Meal timing: what you eat before bed, and when
During my years of high-level competition, I trained from 8pm to 10pm most evenings. Eating at 10:30pm or 11pm was not a choice, it was the schedule. My body had adapted to that rhythm, or so I believed.
Moving away from that structure, I re-learned something the body already knew: when you eat late and go to sleep shortly after, there is a specific kind of discomfort that sets in. You cannot sleep. You cannot do much else either. You are waiting for the digestion to run its course. That waiting time costs you the early part of the night, the part where your body does the most productive recovery work. When you finally fall asleep, the night is already shorter, and the quality of the first hour or two is already compromised. The next morning, the variability score reflects both.
What the science says
The direct research on meal timing and HRV is more limited than for the other levers on this list, and most accessible data comes from practitioner observational datasets rather than controlled peer-reviewed studies. The mechanistic picture is clear: digestion requires autonomic activation for blood flow redirection and gastrointestinal function, which competes with the parasympathetic dominance required for high nocturnal variability. The effect is most pronounced with large, high-carbohydrate meals taken within one to two hours of sleep, where the combined glycemic and insulin response adds a metabolic load on top of the digestive one.
Across wearable observational datasets, the pattern is consistent: meals within two hours of sleep reduce overnight variability in a measurable proportion of users. The magnitude is smaller than alcohol or significant sleep deprivation, but it is controllable. Unlike training load, it has no adaptive benefit. It is pure cost.
The rule that counts
Allow two to three hours between your last meal and sleep. If post-training nutrition forces a late eating window, prioritise protein and fat over a large carbohydrate load. The late-carbohydrate combination is what most reliably suppresses the overnight reading. The meal is not optional. The timing is adjustable.
Hydration: the mechanical lever
Hydration’s effect on variability is mechanical, not neurological. When blood volume drops from fluid loss, your heart rate rises at rest to maintain output. A faster baseline heart rate has less room to vary between beats. The variability score compresses before your nervous system is involved at all.
Training at high sweat output makes the pattern recognisable before any device confirms it: dehydrated sessions feel heavier, RPE is higher for the same pace, and the body is working harder to do the same thing. The variability reading the following morning reflects that mechanical cost, and it resolves with fluid, not rest.
What the science says
The ACSM position stand on exercise and fluid replacement (Sawka et al., 2007) documents the physiological cascade: reduced plasma volume, increased cardiovascular strain for identical absolute workloads, elevated resting heart rate [6]. At the level of HRV, dehydration produces a higher baseline heart rate even at rest. A faster, less flexible heart is mechanically constrained in how much it can vary between beats. The variability score narrows not because of nervous system suppression but because the range of possible beat intervals has compressed.
The most relevant context for HYROX athletes is the morning after a heavy session without full rehydration. A large sweat loss carried into the night, without adequate fluid and sodium replacement, can produce a suppressed variability score the following morning that resolves completely with rehydration. Rest is not the solution. Fluid is.
The rule that counts
If your variability score is low and your urine is dark, rehydrate before drawing conclusions. Check the previous day’s fluid intake and sweat output. A mechanically low reading caused by dehydration does not require training modification. It requires water and sodium.
Why these six levers interact
The harder problem is not understanding each lever in isolation. It is reading a single low number and not knowing which one caused it.
The most common failure is attributing a compound reading to a single cause, usually training load, and reducing sessions as a result. The training was not the problem. Three controllable inputs were. Reducing sessions cost a week of adaptation without solving anything.
Annotating the number matters: noting alcohol, sleep quality, work stress, and meal timing alongside the reading each day is what converts HRV from a stress indicator into a decision tool. The number alone is not enough. The number in context is.
The six checkpoints before interpreting a low reading
Before adjusting your training plan based on a suppressed variability score, run through this in order:
- Alcohol in the last 48 hours? The most reliable and most underestimated suppressor. Rule it out first.
- How was the sleep? Check your sleep score alongside the variability reading. A short or fragmented night is its own explanation.
- Where are you in your training load? A single low day 12 to 24 hours after a hard session is normal. A low weekly average across a full block is different.
- What was your non-training stress this week? Work load, travel, emotional stress. It counts toward total load.
- Did you eat late or heavily before bed? The night before, specifically the last two hours before sleep.
- Were you dehydrated? Previous day’s session, sweat output, urine colour this morning.
If two or more of these apply, your training load is probably not the primary cause. Fix the inputs before adjusting the programme.
What this article does not cover, and what comes next
Knowing what moves your variability score is step one. The harder problem is building a reading protocol that makes the number reliable: how often to measure, under what conditions, and how to distinguish a meaningful multi-day trend from noise produced by a single unusual night.
That is the subject of the next article in the HRV series: How to read your HRV data: a practical guide for HYROX athletes. It covers measurement consistency, the weekly trend framework, and the orthostatic test that turns a single variability score into a readiness decision with much higher confidence. I am currently running that protocol with a chest strap and Elite HRV, and I will publish it with my data once the testing period is complete.
Personal data from my current testing will be added to this article as the protocol progresses.
Does alcohol really affect HRV that much, even after just one or two drinks?
Yes. The effect is consistent enough that it shows up across population-level data, not just in controlled lab studies. Pietilä et al. (2018) tracked over 4,000 subjects across thousands of nights and found that even one to two standard drinks produced a statistically significant drop in overnight variability. The mechanism is direct: alcohol promotes sympathetic activation during the recovery phase of sleep, compressing the parasympathetic window that produces high variability. For athletes who rarely drink, the effect can be particularly visible, there is no built-in tolerance, and the contrast against a stable personal baseline is sharp. Do not use a post-drinking variability score to evaluate training readiness. The number is measuring the alcohol.
How do I know if my low HRV is from training load or from something else?
Context is the only way. A training-load explanation requires that the low reading follows a hard session within the expected 24 to 72 hour window, that sleep was normal, that alcohol was not a factor, and that the reading returns to your normal range by day three or four. If any of those conditions are not met, work through the other five levers before concluding the training load is the problem. Logging a simple daily note (sleep quality, alcohol, meal timing, and perceived stress) alongside the number takes 30 seconds and makes the interpretation considerably more reliable over time.
Does psychological stress really move HRV as much as training does?
For most athletes, not as much as a hard interval session, but more than expected. Your nervous system does not distinguish stress by source. Research from Thayer et al. confirms that emotional and cognitive load produces measurable reductions in parasympathetic activity through the same autonomic pathways as physical training stress. The practical implication for amateur athletes training alongside full-time professional work: a genuinely heavy work period can account for a multi-day suppression in variability that has nothing to do with your training programme. Adjusting your sessions in response to that is usually the wrong call. Adjusting your recovery inputs (sleep, nutrition, rest) is usually the right one.