Stress, and how I felt.

Why it's even worth asking

Start with what the watch is reading. The score it labels 'stress' isn't a mood; it's a read of the autonomic nervous system, the automatic controller that speeds the heart and slows it without anyone asking. In a healthy body that system lifts under effort and settles with rest; from the tiny timing differences between heartbeats, the watch estimates where it sits, minute to minute.

In the illnesses this guide is written for (ME/CFS, Long COVID, the wider post-infectious family) that controller is measurably off. Pooling sixty-four studies, one review found these bodies tend to sit tilted toward 'activated': a faster resting heart rate, less beat-to-beat variability, a system slower to come back down. So the autonomic signal here is not noise. In this population it is known to carry real information about the state of the body.

What almost no one has checked is whether a consumer watch's everyday numbers track any of that in ordinary life, day to day, against how a person actually feels. Wearable research lives mostly in the pacing corner, not here. Which is what makes four years of my own data unusual: I logged it before I ever found the guide, so it was never shaped by knowing what to look for. Whatever it shows, it isn't me finding what I went looking for.

Going deeper: what 'stress' really measures · what this watch can't see · the studies behind this

What shape would we even expect?

Suppose the watch is reading something real. What kind of relationship would it even make against how I felt? The obvious guess is a straight line, more stress, a worse day. But a straight line is probably the wrong thing to expect. Four independent reasons, from published work on how bodies signal, point to a curve instead, and they do it before anyone looks at my data:

1. The number is built from heartbeat timing, and that maths is already curved

   The watch's 'stress' is derived from heart-rate variability, which doesn't track the nervous system in a straight line even in principle. A straight-line comparison is fighting the measure's own shape before any feeling enters the picture.

2. More activation helps, until it doesn't

   The oldest, best-established curve in arousal science: a little activation sharpens you, too much degrades you. Bodies and brains run on an arched curve, not a ramp, so the response needn't climb in step with the load.

3. At the worst end, the body can stop being able to react

   When the system is depleted enough, it can no longer mount the response the watch measures. That means the relationship needn't be monotone at all: both ends of the range can mark trouble, and the curve can run in a direction the obvious story wouldn't guess.

4. The control system is a feedback loop, not a dial

   The brain–heart circuit behind all of this works by inhibition and feedback; when it frays it doesn't fade gently, it can tip. That is non-linear by its very nature.

None of this proves a curve in my numbers; these come from other people's bodies and other cohorts, so they license only the expectation, not the result. But they turn a curve from a special excuse into the ordinary thing to look for. And my own prediction going in? None: I wasn't watching for this when I logged it. The lay guess anyone reaches for is the straight line, which is exactly the shape the science says not to expect.

Why a curve is expected, autonomic-shape literature brief ↗

Why I look at stress, activity and recovery so closely

There's a second reason I examine these numbers in such fine detail, and it's the reason this site exists at all. Wiggers' guide makes specific, testable claims about exactly them. She describes a stress score whose cost isn't even: a step from 30 to 40, she writes, 'costs more than it looks'. She describes stress that, after you overdo it, won't come back down during rest, staying high through the evening, the 'walls of orange'. And she joins the two through activity: when your stress stays stuck up there, she writes, 'the activity you did was probably too disruptive for how you were feeling at the time.'

So her patterns point straight at stress, activity and recovery, read finely, within a single day, and across the run of days around a crash. That is why I examine these channels the way I do, minute by minute and bout by bout. Not, on this page, to grade her (this is still only a description of what the numbers do) but to show why they earn this much attention before any verdict is reached.

See her patterns, held against my data in the field guide.

A first look: the simple test comes up flat

The obvious thing to do is correlate the daily numbers: each day's watch-stress against that day's felt-state. Do it, and the strongest channel barely moves the needle, and most don't move at all. On a straight-line measure, the watch and the feeling look almost unrelated.

A reminder on the word: the watch's 'stress' is an HRV-derived read of how activated my nervous system was, not a measure of mental stress. Lining it up against how I felt is matching an arousal proxy to a feeling, not two feelings.

Not a straight line

A straight-line correlation can only see a straight-line relationship; if the real shape bends, the measure cancels it toward zero. Theory said expect a bend, and a bend is what's here. Look at how the felt-state sits across the watch's stress range and it doesn't climb or fall in a straight line: my best-feeling days sit in the middle of the range, not at the bottom, with the days at either end a little worse. A curve like that is close to invisible to the simple correlation, which is exactly why the simple version reads flat.

Why the middle? One rival I haven't ruled out

A curve was the expected shape, but those reasons predict that the relationship bends, not which way it bends or why. The most everyday explanation for this particular hump is an interaction with activity. On a better day I tend to do more, which lifts the watch's activation off the floor and into the middle of its range; a shut-down day barely moves it; and the very highest readings can come from either real overexertion or a body quietly fighting something off. If that's the engine, then 'best in the middle' would be partly a map of how much I did, not a direct law connecting stress to how I felt. And this isn't a stray worry: it's close to something Wiggers says herself; that when your stress stays high, 'the activity you did was probably too disruptive for how you were feeling at the time.' Her own guide already couples activity and stress, which is what makes this a question her patterns can be tested on, not one beyond them.

And it lives finer than a daily average

The other place the signal hides is time. Instead of one number per day, line up the days around each crash and compare them to matched ordinary days. Now the autonomic signals do show themselves: across the four days before a crash, several run higher than normal, resting heart rate most of all. The signal that vanished in the daily correlation reappears the moment you stop averaging it away.

A description, not a verdict. That these signals ran higher in the days before a crash is what the numbers did; it does not make any of them a reliable warning. Whether one earns that is a separate, stricter test, told in the field guide and a discriminator, not a predictor; some of them did not survive it (what didn't survive).

Why this matters for everything else here

Two standing principles come out of this. A flat correlation is not the same as no relationship; it can mean you're measuring the wrong shape at the wrong scale. And a curve is the normal shape for this kind of signal, not special pleading invented after the fact: it's what the physiology predicts. That is the whole reason the findings on this site lean on event-matched, bout-level, and phase-by-phase methods instead of a tidy daily correlation: a real signal can be there and still be cancelled to zero by a measure too blunt to see its shape. It's also a standing caution, for me, and for anyone who glances at their own watch and feels nothing lines up.

Related: a discriminator, not a predictor, and where the watch agrees with the life I lived, the recovery, in six phases.