Adaptive TDEE: working out what you burn from your own data

Your total daily energy expenditure is the number every calorie target is built on. Nearly every app estimates it once, from a formula, and never checks whether it was right. There is a better way, and it needs nothing you are not already doing.

What TDEE is, briefly

Total daily energy expenditure is everything you burn in a day: keeping you alive at rest, digesting food, deliberate exercise, and all the incidental movement in between. Eat below it and you lose weight. Eat above it and you gain. Every calorie target is just this number with an offset applied.

So if the number is wrong, the target is wrong, and no amount of careful logging afterwards rescues it. You can weigh every gram of food perfectly and still be aiming at the wrong figure.

The usual approach, and why it drifts

The standard method is a prediction equation — Mifflin-St Jeor, or Harris-Benedict — fed your height, weight, age and sex, then multiplied by an activity factor you pick from a dropdown. It gives you a number instantly, which is its whole appeal.

The trouble is that it is a population average dressed up as a personal measurement, and it is wrong for most individuals by a margin large enough to matter. We wrote about exactly how wrong, and why, on a separate page about TDEE calculator accuracy.

The adaptive approach: measure it instead

There is a much better source of truth sitting in data you already generate. Energy balance is not a theory, it is arithmetic:

You already know what you ate, because you logged it. You already know how your weight moved, because you weighed in. So you can solve for expenditure directly rather than predicting it. No equation, no activity dropdown, no guessing whether you are “moderately active”.

Roughly 7,700 calories corresponds to a kilogram of body mass. If you averaged 2,400 calories a day over three weeks and lost 0.6 kg, that deficit works out to about 220 calories a day, which puts your expenditure near 2,620 — regardless of what any calculator said.

The catch, and how it is handled

Body weight is noisy. Water, sodium, glycogen, digestion and the time of day move the scale by a kilogram or more without a gram of fat changing hands. Take two weigh-ins a fortnight apart and subtract, and you are mostly measuring how salty last night's dinner was.

The fix is to stop reading individual weigh-ins and fit a trend through all of them. Daily noise is large but roughly symmetrical, so across enough days it cancels, and the slope of the trend is a far better estimate of what your body actually did than any two points.

This is what Setpoint does with your logs:

1. Fit a trend line through your weigh-ins, rather than comparing two of them.
2. Compare that trend against your logged intake over the same window.
3. Read expenditure off the energy balance.
4. State how confident that figure is, based on how much data went into it.
5. Blend toward the measured number as confidence rises, instead of switching to it.

That last step matters more than it sounds. A target that lurches 400 calories overnight because you had a big weekend is not a better target — it is a worse one, and it destroys any trust you had in the app. So the formula estimate is where you start, and it gives way gradually as real evidence accumulates.

How long before it knows

Realistically a couple of weeks of consistent logging and weighing before the measured figure is worth much, and it keeps improving after that. There is no way around this: the method needs your weight to move enough to be distinguishable from noise, and that takes time.

This is the honest trade. A calculator gives you a number in ten seconds that might be 400 calories out. This gives you nothing new for a fortnight and then gives you something real.

Why it matters most around week six

Almost every diet stalls eventually, and the reason is usually misread. Two separate things are happening at once:

• You are smaller. A lighter body genuinely costs less to run. This part is arithmetic and entirely expected.
• Metabolic adaptation. Beyond the mass change, expenditure tends to fall further — less incidental movement, quieter thermogenesis. You move less without deciding to.

A static TDEE figure cannot see either. It was set on day one and it is still the day one number, so the app keeps telling you the same target while your actual expenditure has moved underneath it.

Measuring expenditure repeatedly separates the two: subtract the part explained by being lighter, and what remains is adaptation. That distinction decides what you should actually do, and the answer is frequently the opposite of eating less.

The short version

A formula predicts your expenditure from people who resemble you. Adaptive TDEE measures it from you. The first is instant and frequently wrong; the second takes a fortnight and is about you specifically.

See how Setpoint works