Should Your HRV Decide Today's Workout?

A kind of morning exists now that did not exist thirty years ago. You wake, and before your feet reach the floor you reach for a number. Two digits, computed while you were unconscious, delivered with the quiet confidence of a verdict. Green: go. Amber: perhaps not today.

I have no objection to the number. I have an objection to the surrender.

The morning number that wants to run your day

The idea underneath the score is not silly. Heart rate variability — the beat-to-beat wobble in the gap between heartbeats — tracks the vagal side of your autonomic nervous system, and that system genuinely responds to accumulated training and life stress. Nothing controversial there.

The controversial part is the leap: from this correlates with fatigue to this should decide your Tuesday. That leap has now been tested in real training studies, on real athletes, for real weeks. The answer is more interesting than either the enthusiasts or the cynics were expecting, and the most useful finding is only a year old. If you want the physiology first, we covered what HRV is and what actually moves it separately.

What HRV-guided training actually is

It is not a recovery score. It is a prescription rule, and the distinction matters.

You measure HRV every morning under identical conditions. You spend a week or two building a personal baseline — a rolling average and a normal band of noise around it. Then the rule runs: if this morning sits inside or above your band, today's hard session goes ahead. If it sits below, the hard session becomes an easy one, or nothing.

Your weekly plan stops being a plan and becomes a queue. The hard work still happens; it happens on the days your physiology says it can be absorbed. Nobody trains less on average. They train hard on different days than they intended to.

That is the whole mechanism. It decides when, not what — which is less exotic than the marketing, and also the reason it works at all.

The evidence is real, and smaller than the marketing

Six randomised trials have pitted this rule against a fixed, predefined plan in endurance athletes. A 2020 systematic review and meta-analysis pooled them: 195 participants, 134 men and 61 women, all measured on VO₂max before and after.

Both arms improved — training works, which is not news. The guided arm improved more: an effect size of 0.402 (95% CI 0.273 to 0.531) against 0.215 (0.101 to 0.329) for the fixed plans, with the between-group difference significant at p<0.0001. The advantage was larger in amateurs than in elites, and larger in women than in men.

Read that honestly and it is worth having, not transformative. And the reason it works is deflating in the best way: nothing about consulting a number builds a mitochondrion. What changes is allocation. The rule blocks the two ordinary errors that quietly ruin ordinary training — doing the hard session on a day you cannot absorb it, and coasting through an easy day when you had more in you. The gain is scheduling, not physiology.

The authors flagged something quieter and more interesting than the headline effect: the individualised arm produced fewer non-responders. Not a bigger average. A more reliable one.

The 2025 trial where the number alone did nothing

Then, in September 2025, a trial in Scientific Reports took the rule apart to see which piece was doing the work.

A hundred and nineteen cyclists signed up. Twenty-eight finished — which is itself a finding about how much daily self-measurement people will actually tolerate. Those who lasted were serious: mean age 45.6, eighteen to twenty-one years of riding, nine to eleven hours a week. Forty days total, nine of them spent establishing baselines and thirty-one taking daily instructions from an algorithm.

Three versions of that algorithm. Group one let HRV decide alone. Group two gave HRV a co-signer: a short morning question about sleep, fatigue, soreness and stress. Group three added resting heart rate on top.

Pooled across all twenty-eight riders, everything improved. Split by group, the picture changes completely. Group one — the pure number, the version most apps actually sell — showed no significant within-group improvement on any power measure. Group two gained on one-minute and five-minute power. Group three gained on five-minute, twenty-minute and functional threshold power, the measures that matter most to a cyclist.

Eight people per arm. No arm trained without any guidance at all, so this compares flavours of the algorithm rather than algorithm against plan. Men only. Four different measuring devices between them. This is not proof of anything, and the authors say so first. It is a direction — and it happens to be the direction a decade of monitoring research already pointed.

Why a free question beat an expensive sensor

A 2016 systematic review in the British Journal of Sports Medicine went through the athlete-monitoring literature and reached a conclusion that should have been more embarrassing to the industry than it was: subjective self-reported measures showed superior sensitivity and consistency in reflecting training load compared with the objective ones — hormonal markers, performance tests, the expensive things. Wellbeing fell when acute load rose. It fell further as chronic load accumulated. It recovered when the load came off.

The most sensitive instrument in sports science turns out to be a person answering four questions honestly. It costs nothing, and nobody has worked out how to sell it to you.

The reason is not mystical. A sensor measures one axis of one system. The question integrates your entire life — the argument last night, the deadline, the second glass of wine, the knee that has been quietly complaining for a week. HRV is genuinely blind to most of that. It knows your vagus nerve; it does not know your Tuesday.

One honest caveat, because it matters: self-report degrades the moment you have a reason to shade it. An athlete who reports fatigue to a coach who picks the team learns to report less of it. Alone in your own bedroom, answering to nobody, you have no incentive to lie — which is precisely the setting where the cheap instrument works best.

One reading is weather; the band is climate

Eleven elite water polo players wore a wrist device nightly for the sixteen weeks before the Tokyo Olympics. Their weekly coefficient of variation for HRV averaged 5.4 ± 0.7%, and for resting heart rate 7.6 ± 1.3% — sitting comfortably inside the 3–13% range established by other measurement methods.

That is the sentence to keep. Bouncing around is not a malfunction; it is what a healthy, well-trained autonomic system does. The authors' practical conclusion was that shifts of roughly 10–45% clear the noise floor and can be trusted. Smaller wobbles cannot.

So today-versus-yesterday is a comparison of two noisy draws, and it will hand you a different answer every day of the week. Your seven-day rolling average against your own baseline band is the comparison that carries information. Resting heart rate behaves the same way and is worth reading alongside it — we went through what a resting heart rate actually tells you, and more broadly, what a wearable can and cannot measure.

One more thing worth not panicking about: a low morning after a genuinely hard session is not a warning. It is the receipt. The signal worth acting on is a low that persists after the load has come off.

How to run this yourself for two weeks

None of the trials above used anything you cannot do with a phone and some discipline. If you want to try the rule rather than read about it, this is the honest version.

Fix the measurement before you trust it. Same time, same posture, before your phone and before coffee. Consistency of method matters more than which device you own.

Earn a baseline before you obey anything. Measure for one to two weeks and change nothing. You are looking for your own average and your own normal spread. A number without a band is just a number.

Decide with two inputs, never one. The reading, plus one honest answer on sleep, soreness, stress and mood. That is the finding of the 2025 trial reduced to a habit.

Let it veto the intensity, not the week. A below-band morning turns a hard session easy. It does not turn a training week into a rest week. If you need a real rest week, plan one on purpose — we looked at what the deload evidence actually shows.

And keep the whole thing in proportion. HRV, resting heart rate and sleep figures in the Agen app are wellness trends to notice over weeks, not medical measurements and not a diagnosis. If a low trend persists alongside symptoms that concern you, that is a conversation with a doctor, not with an app. For the wider question of when to push and when to hold, we wrote a whole piece on reading your own recovery.

The bottom line

HRV-guided training works, modestly, and for a boring reason: the rule is a better scheduler than your ambition is. It moves the hard work onto the days that can hold it.

But the 2025 evidence suggests the number is not the active ingredient by itself. What worked was the number plus a person willing to answer honestly about their own morning — and the version with no human vote at all was the version that went nowhere. Use numbers to correct fantasy, not to replace experience.

When the app and the body disagree at seven in the morning, remember which of the two has to do the session.