Your Recovery Score Is Not a Verdict

You wake up, and before your feet are on the floor there is a number waiting. Recovery: 41 per cent. Strain is elevated, sleep was fragmented, the suggestion on the screen is a light session or a rest day. Some platforms have stopped suggesting and started rewriting — the heavy squats you had planned quietly become mobility and an easy walk.

The number is not made up. Heart rate variability is a real signal, it does track systemic load, and it will occasionally catch something before you do: the illness two days out, the week of poor sleep you had stopped noticing, the training block that has tipped from hard into too much.

So this is not an argument that the data is wrong. It is an argument about who is holding the pen.

A plan you renegotiate every morning is not a plan. It is a daily referendum, and referendums held at six a.m. do not go well for the person who has to run.

The problem is the frequency

Consider what has actually changed. Not the existence of bad days — those were always there, and every honest training programme accounts for them. What has changed is that there is now a formal, numerical, apparently objective invitation to skip, and it arrives every single morning without being asked for.

Most mornings before a hard session feel bad. That is not a malfunction; it is what the morning before a hard session feels like. Previously this was ambient noise and you went anyway, because the alternative was admitting you had decided not to on the basis of a feeling. Now the feeling has a number attached, and the number came from a device, and skipping stops being a decision you made and becomes a recommendation you followed.

That is a meaningful difference. It is much harder to notice a pattern in yourself when every instance of it was externally justified.

And the sensitivity runs the wrong way for the failure mode most people actually have. Genuine overtraining is rare in anyone who is not an athlete. Under-training, drift and quiet attrition are extremely common. A tool that is quick to say not today is well calibrated for a problem you probably do not have, and poorly calibrated for the one you do.

Veto, not vote

The arrangement that works is not to ignore the score. It is to decide in advance what the score is allowed to do.

Set the week's sessions on Sunday, when you are neither tired nor about to train, and treat that as the plan of record. Then give the number one specific power: it may veto, and only on a threshold you defined before you saw it. Two consecutive terrible nights. A resting heart rate well outside your normal range. Anything alongside a sore throat. Those are real, and stopping is correct.

What it does not get is a vote on an ordinary Tuesday when it says 58 and you would rather not.

The distinction matters because a veto is rare by construction and a vote is available daily. One protects you from the small number of days when training is genuinely a bad idea. The other slowly converts your training into something you do when conditions are favourable, and conditions are favourable about a third of the time.

There is a modest, unfashionable version of all this that keeps working. Look at the number after the session rather than before it. Let it tell you what happened rather than what is permitted. Over a month it will teach you something honest about how you respond to load, which is what it is genuinely good for.

The device knows a great deal about your night. It knows nothing whatsoever about your February, and February is where this is decided.