Why Your Watch's Calorie Burn Disagrees With Your TDEE

You put your details into a TDEE calculator and it tells you that you burn around 2,400 calories a day. Your watch, which has been on your wrist the whole time and has actual data about your heart rate and movement, says 2,900. The watch feels more credible — it was there.

It usually is not. Wrist devices measure heart rate reasonably well and estimate energy expenditure poorly, and the gap between those two facts is the source of most of the confusion.

This is not a small technical quibble. A 500-calorie disagreement is exactly the size of the deficit most people are trying to create, so getting it wrong does not slightly slow your progress — it cancels it entirely.

The most-cited evaluation came out of Stanford in 2017. Researchers tested seven wrist-worn devices against clinical-grade reference measurements — indirect calorimetry for energy expenditure, ECG for heart rate — across sixty participants during walking, running and cycling.

The heart rate results were reassuring. Most devices measured heart rate with a median error under five per cent, which is genuinely good for optical sensing at the wrist.

The energy expenditure results were not. Not one device achieved a median error below twenty per cent. The best managed just over that; the worst was near ninety per cent. And crucially, the errors were not consistent — a device that overestimated one person's burn might underestimate another's, which means you cannot even correct for it with a personal fudge factor.

The reason is that the device is not measuring energy at all. It is inferring it from heart rate, movement and your entered height, weight, age and sex, through a proprietary model that was fitted to a population you may not resemble. Heart rate is a poor proxy for energy cost in the first place — caffeine, heat, stress, dehydration and sleep all move it independently of how hard you are working.

Beyond the raw accuracy problem, there are specific and predictable ways the number inflates.

It would be dishonest to present the calculator as the truth and the watch as the error. A TDEE calculator is also an estimate, and it has its own weak point.

The BMR half is well validated. The Mifflin-St Jeor equation was derived from indirect calorimetry on nearly 500 adults, and a systematic review found it predicts resting rate within ten per cent for around seventy per cent of people. That is a genuine measurement-based foundation.

The activity multiplier is where the judgement lives. Choosing between 'lightly active' and 'moderately active' changes the answer by several hundred calories, and most people place themselves too high — training hard four times a week while sitting for the other twenty-three hours a day is closer to lightly active than very active.

So the honest framing is: the calculator's uncertainty is concentrated in one input you control and can reason about, while the watch's uncertainty is distributed through a proprietary model you cannot inspect. That is why the calculator is the better starting point, not because it is more accurate in principle.

There is a third factor that makes both numbers optimistic, and it is the most interesting one.

Herman Pontzer and colleagues have found that total daily energy expenditure does not rise in simple proportion to activity. Comparing populations across a wide range of physical activity levels, including hunter-gatherer groups whose daily movement far exceeds a Western office worker's, differences in total daily burn were much smaller than the differences in activity would predict.

The interpretation is that the body compensates. Increase activity and some of the extra cost is offset by reductions elsewhere — less spontaneous movement, lower resting expenditure, reduced spending on other physiological processes. How large and universal this effect is remains debated, and the strong version of the claim is not settled.

What is not debated is the practical implication, which everyone who has tried to out-train a diet already suspects: the calories you burn in a session do not all show up as extra calories burned that day.

The solution is not to find a more accurate number. It is to stop needing one.

If the numbers still do not add up after a few weeks, the intake side is worth examining before the expenditure side.

Studies using doubly labelled water — the reference method for measuring energy expenditure in free-living people — have repeatedly found that people under-report what they eat by twenty to forty per cent, without any intention to deceive. Portion sizes drift, cooking oil goes uncounted, drinks and tastes-while-cooking vanish from the record.

Weighing everything for a single week is unglamorous and settles the question faster than any amount of adjusting your activity multiplier. More often than not, the expenditure estimate was fine.