HRV for hybrid athletes: how to read it without overreacting

There's no "good" HRV number for athletes that's worth chasing. HRV varies so much between people and devices that the only comparison that means anything is your own 7-day average against your own recent baseline. Single mornings are mostly noise. A shift that holds for several days, alongside a rising resting heart rate and sessions that feel worse than they should, is the thing to act on.

This is a training signal, not a health check. If you feel ill, have chest pain, faint or notice anything that worries you, stop and see a clinician. A watch reading can't rule anything in or out.

What 48 days of my HRV looks like#

Here's my overnight HRV from 2 June to 19 July 2026, as my Garmin recorded it, one reading a night, with the trailing 7-day average on top. The daily number ran anywhere from 58 to 147 ms. The 7-day average stayed between 81 and 128.

Ken's overnight HRV: daily vs 7-day average

day (day 1 = 2 June 2026)	Daily (ms)	7-day average (ms)
1	60
2	77
3	79
4	80
5	88
6	83
7	118	83.6
8	78	86.1
9	82	86.9
10	102	90.1
11	111	94.6
12	63	91
13	85	91.3
14	83	86.3
15	89	87.9
16	87	88.6
17	105	89
18	98	87.1
19	72	88.4
20	103	91
21	119	96.1
22	145	104.1
23	147	112.7
24	122	115.1
25	132	120
26	127	127.9
27	95	126.7
28	100	124
29	82	115
30	82	105.7
31	58	96.6
32	98	91.7
33	84	85.6
34	76	82.9
35	87	81
36	91	82.3
37	89	83.3
38	80	86.4
39	94	85.9
40	105	88.9
41	66	87.4
42	97	88.9
43	100	90.1
44	110	93.1
45	67	91.3
46	113	94
47	103	93.7
48	107	99.6

Source: Ken's own Garmin readings in hybridAF, 2 June to 19 July 2026, read from the database on 22 October 2026. The 7-day average starts on day 7.

Look at 12 to 13 June. My HRV dropped from 111 to 63 ms overnight, a 48 ms fall. Treat that morning as a verdict and you'd pull the session. The 7-day average barely moved, from 94.6 to 91.0. That's the whole argument of this article in one data point.

It's also normal. Buchheit's review puts the day-to-day coefficient of variation of the standard research measure (log-transformed rMSSD) at 10 to 20% in athletes, driven by ordinary changes in training load.¹ Plews and colleagues, writing about elite endurance athletes, recommend averaging rather than single days, and say long-term monitoring is needed to learn each athlete's "unique individual HRV fingerprint".²

Why population "good HRV" charts don't help you#

Search "good HRV for athletes" and you'll find tables by age. They aren't much use to you, for three reasons.

First, the research that's useful for training doesn't compare you with other people. It compares you with you. The HRV-guided training trials set each athlete's thresholds from that athlete's own baseline weeks, not from a population table.³⁴

Second, watches don't measure the same thing. Apple Health stores HRV as SDNN, the standard deviation of the gaps between normal heartbeats, recorded automatically by the watch.⁵ Most training research uses rMSSD, which is built from successive differences between beats and is the index Buchheit recommends for athletes in the field.¹ They're related, but they're not interchangeable, so a number from a paper or a friend's different watch isn't a benchmark for yours.

Third, timing. The research protocol is a 5 to 10 minute reading on waking, same position every day.¹ Apple's documentation doesn't say when the watch takes its samples.⁵ Overnight and morning readings from consumer devices are convenient, but they're not the lab protocol, which is one more reason to stay inside your own data.

How to actually read it#

Use a 7-day rolling average#

Averaging is the single biggest fix. Buchheit notes that noise falls by a factor of 1/√n, so averaging four readings halves it, and that in the studies he reviews, links between HRV and running performance only showed up when at least 3 to 4 days were averaged, not on isolated days.¹ In the triathlete case study, it was the 7-day rolling average that trended down as one athlete slid into non-functional overreaching, while the control athlete's stayed flat.⁶

You don't need every morning#

Plews and colleagues tested how many randomly picked mornings a week you need before the weekly average behaves like the full 7-day one. In trained triathletes going through an overload block, the size of the change they could see climbed until about 3 days, then flattened.⁷

HRV change detected vs mornings averaged per week

	Value (standardized change)
1 day	0.2
2 days	0.33
3 days	0.49
4 days	0.48
5 days	0.47
6 days	0.45
7 days	0.43

Standardized change in Ln rMSSD from baseline to overload, by number of randomly selected days averaged. Source: Plews et al. 2014, IJSPP, abstract. Trained triathletes.

Their conclusion is a minimum of 3 valid readings a week.⁷ That was trained triathletes. If your readings are noisier than theirs, more mornings will help, and a watch that records every night makes it free anyway.

Define "normal" as a band, not a line#

Two of the HRV-guided trials, one in recreational runners and one in trained cyclists, set a band around each athlete's baseline at the mean plus or minus half a standard deviation, and only swapped hard sessions for easy ones when the 7-day average left it.³⁴ Inside the band, treat it as noise. Outside it is something worth a look. Both trials built that band on rMSSD from a dedicated morning reading. Nobody has validated the same half-SD rule on Apple Health's SDNN, so if that's your data, treat the band as a sensible starting rule, not a tested threshold.

Wait for it to persist#

Buchheit's advice is to wait until a change shows up over 2 to 3 consecutive days (or weeks, for weekly averages) before acting.¹ One bad morning after a late night or a hard session is expected. A week of it isn't.

Build the baseline in a normal stretch#

Your band is only as good as the weeks it's built from. The runners in Vesterinen's trial built theirs over 4 weeks of ordinary training, and the authors warn against recording the baseline during heavy stress such as work stress or illness.³ The cyclists in Javaloyes's trial used a 2-week baseline.⁴ If you start tracking during exam season or the week after a race, expect the band to shift once life settles, and don't read much into the first month.

Know what else moves it#

Training isn't the only input. Buchheit lists cases where HRV went up, not down, within a day of heavy training in the heat, despite athletes feeling worse.¹ He also notes that a single ectopic or skipped beat in a 5-minute recording can shift common HRV indices by up to 50%.¹ A watch won't tell you that happened. It just hands you a strange number, which is one more reason the average beats the single reading.

A weekly routine that doesn't take over your life#

Here's one way to turn all of that into something that takes 30 seconds a morning:

The middle two rows follow the logic of the trials and reviews above: act on sustained shifts, read HRV with resting HR and context, and remember that fatigue can push HRV up as well as down.³¹⁸ The exact cut points are a starting rule, not a law.

Low isn't always bad, and high isn't always good#

This is where people overreact in both directions. Plews and colleagues point out that in elite athletes, both increases and decreases in HRV have been tied to negative adaptation.² A meta-analysis of functionally overreached endurance athletes found that their weekly averaged HRV went up, a moderate effect (SMD 0.81, 95% CI 0.35 to 1.26), while single-day values showed nothing reliable.⁸

So a big HRV number after a brutal training week isn't automatically "fully recovered". Buchheit's point is that HRV has to be read alongside resting heart rate and what you've actually been doing; a reading with no training context is likely to mislead you.¹

What a real multi-day shift looks like#

Here's the other kind of drop, from my Apple Health data in September. Those readings are SDNN, so they sit on a different scale from the Garmin numbers above; each is only compared with its own history. From 7 to 10 September my HRV read 49, 25, 34 and 35 ms. My resting heart rate went to 49, 56, 48 and 47 bpm. Against the 28 days before 8 September, my usual resting HR was 43.7 plus or minus 1.9 bpm, so 56 was about 12 beats above my average. HRV was well below its range on every one of those nights.

That's the pattern worth respecting: several days, two signals moving together, both well outside their bands. The data doesn't say why, and it shouldn't pretend to. When that happens, the training answer is easy days or rest until things come back toward normal. If you also feel unwell, that's a question for a clinician, not for a chart.

The part the research hasn't covered: lifting#

Almost everything above comes from endurance athletes: triathletes, runners, cyclists.⁶³⁴⁸ None of these studies set HRV thresholds for people who also squat heavy three times a week. The averaging logic still holds, because noise is noise. What we don't know is how much a heavy lower-body session moves your next-morning HRV compared with an interval session, or whether the same band works when half your week is barbell work.

The practical answer is to keep the method and add a log. If your 7-day average dips, look at what you did that week before you decide it means anything. That's the same advice Buchheit gives for endurance athletes, applied to a mixed week.¹

How hybridAF reads your HRV#

The Body screen shows each night against your own usual range: the average of the 28 days before that reading, plus or minus one standard deviation, once there are 7 readings. The recovery score weights HRV at 45%, resting HR at 30% and sleep at 25%; HRV and resting HR are compared with your own baseline, which is the average of all your stored readings rather than a rolling window, and sleep with a fixed scale. Those weights and that band are our design choices, not published standards. A one-SD band is wider than the half-SD band in the training studies, so it flags fewer nights on purpose. The HRV it reads from Apple Health is SDNN. There's a walkthrough in Reading Body, and the evidence and the design choices are separated on the science page.

Where to go next#

If you want to know whether adjusting training day by day from HRV actually produces better results, read Does HRV-guided training work?. Resting heart rate, the cheaper signal used above as a cross-check, gets its own page in resting heart rate as a training signal. And HRV only tells you about cost; for the work side of the equation, see training load for two sports. Everything on recovery is collected in the recovery hub.