How to read your HRV data: a practical guide for HYROX athletes
You read your variability score to decide whether to push or adapt the session, not to decide whether to skip it. Here is the protocol that turns a morning number into a training decision, and keeps you working.
6 June, just past 6am. I woke up flat. Not a specific ache, not a tight hamstring or a stiff shoulder, just a generalised heaviness sitting across the whole body. And I was not surprised. The night before I had stacked a maximal interval session on the curved treadmill with an all-out circuit in the evening. The bill had come due.
I reached for the watch already knowing roughly what it would say. 21.7 ms. The lowest reading I had ever recorded.
The number was not the interesting part. This was: my plan already had an easy Zone 2 run scheduled for that morning, exactly the session the data would have prescribed if I had been sitting low for three days straight. The plan and the reading pointed to the same place.
That is when it landed for me. A variability score does not only tell you how to train today. It tells you whether your plan was built right in the first place.
If you have read what HRV actually measures and what moves your variability score, you have two pieces of the framework. The first tells you what the number tracks. The second tells you which inputs drive it.
This is the third piece: how to read the number when you see it every morning, and what to do about it.
Here is the part most HRV content gets wrong. A low reading is not permission to skip the session. It is information about what kind of work your body can absorb today. The number does not decide whether you train. It helps you decide how.
That distinction is the whole point. Read without a protocol, your variability score becomes an excuse generator. Read with one, it becomes a tool that lets you keep training hard while training smart.
Your daily reading is not your baseline
The number on your watch this morning is not your reference. Your rolling average is.
A single reading is a data point. Your baseline is the moving average those points orbit. Without it, you cannot know whether today is high, low, or normal for you.
Here is the trap, straight from my data. Across 22 days my raw period average was 77.9 ms. Looks like a clean baseline. But the first five days, coming out of a rested phase, ran 89 to 119 ms and pulled the whole average up.
My real current baseline, the 14-day rolling average as of 12 June, is around 67 ms. Ten milliseconds lower than the period average implied. The wrong reference would have told me every normal day was a “low” day.
What the science says
Plews et al. (2013) showed in elite endurance athletes that the weekly trend, not the daily number, is the unit that tracks training adaptation [1]. Day-to-day swings of 10 to 20 ms are normal and driven by dozens of inputs unrelated to readiness. The signal lives in the rolling average.
The published standard is a 5 to 7 day window. Flatt and Esco (2015) confirmed that a rolling average “provides a more accurate reflection of overall training adaptation” and stops training from being hyper-reactive to single readings [4]. My dashboard runs a 14-day baseline for a more stable reference, then watches the 7-day trend on top of it. Both windows are defensible. What matters is that you track an average, never a single morning.
The rule that counts
Track your rolling average, not today’s number. One reading outside your normal range is noise, not a reason to change anything. A multi-day move in the same direction is where the information starts.
Signal vs noise: the threshold the pros actually use
Not every dip below your baseline means something. Most do not. Your variability score swings 10 to 20 ms between days even under stable conditions. The job is knowing when a move is real.
The professional tool for this is the Smallest Worthwhile Change (SWC). It is the dashed line on every Plews and Buchheit chart: the band around your baseline inside which everything is noise.
Plews and Buchheit define it as 0.5 times the coefficient of variation of your rolling log variability [2]. My dashboard computes it the close-cousin way, 0.5 times the standard deviation of my log variability score. Over my last 14 days that puts my signal band at roughly 55 to 80 ms around a baseline near 67.
Inside the band: noise. Train as planned, no second-guessing. Below the band: a real signal worth a closer look, not an automatic rest day. The band tells you when to investigate. It never tells you to stop.
One day below the band is still not enough to act on. Plews’ principle, refined by Flatt and Esco (2017), is that it takes a trend: multiple consecutive days below baseline, or a falling weekly average, before the number means anything [5]. My dashboard treats one day below baseline as noise and three consecutive days as a real signal.
The rule that counts
Build your own band: 0.5 times the standard deviation of your log variability, or let your app compute the SWC. Below it for one morning, note it and move on. Below it for three mornings, then you investigate the cause. Most mornings, the honest read is: this is noise, do the session.
Why intensity crashed my HRV harder than volume
The most useful finding in 22 days of data: my lowest reading did not follow my heaviest day.
On 30 May I logged 118 training load points, the heaviest day of the block, across intervals and multiple runs. The next morning: 44.3 ms. A clear drop, the kind a big day is supposed to produce.
On 5 June the load was 69 points. Half the volume. The next morning: 21.7 ms. The lowest reading in the entire dataset, more than 20 ms below the 30 May crash.
The difference was the type of work, not the amount. The 5 June session was VO2max intervals (efforts at my aerobic ceiling) on a curved treadmill, plus an evening AMRAP (as many rounds as possible, a timed all-out circuit). Two maximal-output efforts in one day. The 30 May day was high volume at sustained intensity. Different cost. Deeper crash.
This is the core lesson for reading HRV: it is an intensity gauge, not a volume gauge. It tells you how much hard, high-output work your nervous system is carrying, far more than it tells you about kilometres covered. A low reading after a maximal session is the system working, not a warning.
What the science says
Buchheit (2014) separates peripheral fatigue (muscular, metabolic) from central nervous system fatigue in HRV monitoring [3]. Maximal-intensity efforts near the aerobic ceiling produce central fatigue through full neural recruitment, suppressing recovery activity beyond what the raw load number shows.
The practical takeaway is direct. Your training log’s load score is an incomplete proxy for HRV impact. Two short maximal efforts can crash the number harder than a long moderate day. When you read a low morning score, the type of session yesterday matters as much as the total load.
The rule that counts
After any day with two high-intensity efforts, expect a low reading the next morning whatever the load score says. Do not read it as damage. Read it as the cost of the work, and let the 48 to 72 hour window play out before you draw any conclusion. And note the rebound: easy aerobic work often clears it faster than sitting still.
The orthostatic test: from a passive number to an active decision
Your overnight score tells you how you recovered. It does not tell you how your system responds when you actually load it. The orthostatic test does.
The protocol: measure lying down for 5 minutes on waking, then stand and measure for 5 more. The gap between the two is the data point.
Variability holds or rises on standing: your recovery system is handling the challenge. Variability drops hard on standing: your system is already working to manage a simple position change, before you have trained at all.
What the science says
Buchheit (2014) identifies the orthostatic test as one of the most sensitive field tools for separating states that produce identical overnight averages [3]. Overreaching and normal fatigue can look the same on a nocturnal reading. A well-adapted athlete maintains or raises variability when standing; a less recovered one drops. The overnight number catches the result. The orthostatic test catches the mechanism.
The rule that counts
When your overnight reading is below your band, run the orthostatic test before deciding anything. Normal response: do the planned session, start easy, reassess at warm-up. Blunted response (a hard drop on standing): keep the work, change the type, lower the intensity. The number raised the flag. The standing test tells you how loud it is. Neither says stop.
What I actually do with the number
Reading the number is half the job. Acting on it is the half that pays off. As exercise physiologist Andrew Flatt, who has published extensively on athlete HRV, puts it: “interpreting the data is only half the battle. Taking action with appropriate interventions is where everything pays off.” A perfectly interpreted reading you never act on changed nothing.
I do not make the call on the variability score alone. I read it against three things at once: the HRV trend, my training load balance, and how I actually feel. When they disagree, the disagreement is the information. A low score with fresh legs and good sleep is a very different morning from a low score that lines up with heavy load and poor wellness.
That layered read, turning your numbers into a specific session decision, is what I do for athletes one to one. If you want your own data read that way, there is a note at the end of this article.
The 5-step morning reading protocol
A decision framework only works if the conditions are consistent. This takes under 10 minutes.
-
Measure under identical conditions. On waking, before movement, caffeine, or food. Same device, same position, every day. Inconsistent conditions, useless baseline.
-
Compare today against your rolling baseline. Not your all-time average, not your training partner’s number. Is today inside your band, below it, or above?
-
Check the direction of the trend. Is the rolling average holding, rising, or falling over 5 to 7 days? A low day on a rising trend is not a low day on a falling one.
-
Run the orthostatic test if you are below the band. Lying to standing. A normal response points to recent training; a blunted one points to real load.
-
Log one line of context. Yesterday’s session type, sleep, alcohol, stress. The number alone is ambiguous. The number in context is a decision.
Then act. And the action is almost always to train. Here is the decision the reading actually produces:
Want a personalised read of your HRV data?
The framework above is the starting point. Applying it to your data, your sessions, and your training block is a different step. If you want someone to look at your numbers and tell you what they actually mean for how you should train this week, that is exactly what the HRV follow-up service does.
Four places available. Get in touch.
What comes next in the HRV series
This article covers the reading protocol. The next piece completes the picture with personal data:
Cold immersion and HRV: 4 weeks of personal data. Does a 12 degree cold plunge move your variability score measurably, and in which direction? I am running the testing protocol now and will publish the full dataset at the end. Current status: Still testing.
How many days of data do I need before my HRV baseline is reliable?
Most practitioners use a 3 to 4 week minimum. Below that, one unusual week (illness, travel, a peak block) can skew your reference enough to mislead you. Garmin HRV Status uses a 3-week rolling window; Elite HRV recommends 4 to 6 weeks for athletes with variable load. Your first two weeks are directional, not definitive. By week four, with consistent measurement conditions, the rolling average stabilises. Start tracking now. The data only gets more useful with time.
My HRV was high this morning but I felt terrible during the session. What happened?
HRV measures autonomic readiness, not every input that decides session quality. Glycogen depletion, cognitive fatigue, muscular soreness: none show up consistently in a morning reading. The number captures your nervous system’s recovery state, not your fuel stores or leg fatigue. A high score is a green light for your nervous system, not a guarantee of a good day. Use it alongside how you feel at warm-up, then train. The body often outperforms a bad feeling once it is moving.
My variability score has been low for a week. Should I take a rest week?
Not automatically, and rarely. First check the controllable inputs from what moves your variability score: sleep, alcohol, meal timing, work stress. Fix those before you touch training. Second, check your block phase. A loading week is supposed to suppress the trend; the rebound comes in the recovery week. The only version worth acting on is a decline that continues through a recovery week without rebounding. Short of that, keep training and change the stimulus, not stop.