How Is a Readiness Score Calculated

Every morning a ring, a watch or an app hands you a single number and a word. Readiness. Recovery. Body Battery. Almost none of them show you the arithmetic that produced it, and the arithmetic is the part you would actually need in order to do anything about the result.

This page walks through the calculation in the order it happens, then through what the calculation throws away, which is the more useful half.

The inputs almost every score uses

The exact list is proprietary at every vendor, but the ingredients are broadly the same across the category, because they are the signals a wrist or a finger can actually record overnight.

Heart-rate variability, usually averaged across a window of your deepest sleep rather than the whole night. Resting heart rate. Respiratory rate. Sleep duration, and some representation of sleep stages. Skin or peripheral temperature, expressed as a deviation rather than an absolute. Some measure of the previous day's physical load. Often, how long it took your heart rate to fall after you went to bed.

Seven or eight numbers, give or take. That is the raw material.

The three steps that turn those into one number

Step one, each input is scored against a reference. This is the step that decides almost everything downstream, and it is the step nobody shows you. Your HRV of 48 is not meaningful on its own. It becomes meaningful only when compared against something: your own trailing average, your own range for this weekday, or a population band for your age and sex. Which reference gets used is a design decision, and different vendors make it differently.

Step two, each scored input is weighted. The weights are not equal. In most consumer systems HRV and resting heart rate carry the heaviest load, sleep duration a moderate one, temperature a smaller one that spikes in influence when it moves far enough. Some systems change the weights depending on context, so a temperature deviation that would be ignored in isolation becomes dominant when it arrives alongside a raised resting heart rate.

Step three, the weighted parts are combined and mapped onto a scale. Usually zero to one hundred, occasionally a set of named bands. The mapping is often non-linear, which is why the difference between 55 and 65 can feel larger than the difference between 80 and 90.

What the compression discards

At the end of step three you hold one number. Three things did not survive the journey.

The inputs. You cannot see which signals moved, or by how much.

The weights. You cannot see which of those movements mattered to the result and which were nearly irrelevant.

The reference. You cannot see what the comparison was made against, which means you cannot tell whether a low score means you deviated from yourself or merely differ from the average person your age.

This is the structural point, and it is not a criticism of any particular vendor. A composite score is a compression, and compression is lossy by construction. The number is a conclusion. The reasoning behind it was removed before the number reached you.

That is why a score is hard to act on. A conclusion you cannot inspect gives you no purchase. It tells you that something moved. It does not tell you what.

Why two devices disagree about the same night

Once you can see the three steps, disagreement stops being mysterious.

Two systems can record almost identical raw signals and still produce scores twenty points apart, because they sampled HRV over different windows, compared it against different references, and weighted it differently against sleep. Neither is malfunctioning. They answered slightly different questions and then rounded both answers to one number each.

The gap between two devices is often more informative than either number alone, because the gap points at which input the two systems disagree about.

Reconstructing the reason by hand

Until a system shows you the reasoning, this is the manual version. It takes about five minutes.

Open the detail view rather than the score, on whatever device you have. Almost all of them expose the component metrics somewhere below the headline number.

Write down the three or four components and, for each, how it compares with your own recent typical value. Not the population band, your own last fortnight.

Look for the one that has moved furthest from your normal. That is usually the driver, because the heavily weighted inputs are also the ones that move the score when they deviate.

Then look one to three days upstream of it for an ordinary cause. A late or heavy meal. A warm room. Alcohol. A short night before the night being scored. Travel. Illness beginning. Training load from earlier in the week that never cleared. The cause is usually ordinary, it usually precedes the score by more than a day, and it is usually not recorded anywhere the device can see.

If nothing has moved much, the score change is probably noise, and the correct response is to do nothing and wait for the run rather than the night.

What a readiness score cannot tell you

It cannot tell you why. That information was discarded in step three.

It cannot tell you whether the change is meaningful for you specifically, unless you know which reference it used.

It cannot tell you anything clinical. A score is a wellness signal computed from consumer sensors. It does not detect illness, and a number moving is not a finding about disease.

How NuVARD approaches the same problem

The position we take is that the reason should arrive attached to the number, not be reconstructed by the reader afterwards.

Signals are read against your own history rather than a population range, so a deviation means a deviation from you. Each signal is also read against the others, which is what makes it possible to say which input moved first and which followed. NuVARD connects 300+ devices and apps, and holds 53 physiological variables across 15 body systems, each with its own estimate, baseline, trend and levers.

Confidence is stated in plain language, early signal, emerging pattern or strong evidence, rather than as a percentage, because a percentage implies a precision that consumer sensors and a short history do not support.

And every forecast ends in one of three states: held, still open, or missed. The missed ones are published alongside the rest, which is the only mechanism we know of that lets you calibrate how much weight the next statement deserves.

The short version

A readiness score is built by scoring several overnight signals against a reference, weighting them, and compressing the result into one number. The inputs, the weights and the reference are all discarded in that final step, which is why the number tells you that something changed and never what.

Read the components rather than the headline. Compare them with your own recent history rather than a population band. Look one to three days upstream for an ordinary cause. And treat any system that will not show you its reasoning as giving you a conclusion you are expected to take on trust.

NuVARD is wellness and health intelligence. It does not diagnose, treat, cure or prevent disease, and nothing here is medical advice.