Are the calories-burned numbers worth anything?

The number is confident, specific and displayed prominently: 612 calories. It is an estimate produced by an equation, and it is probably wrong by more than a meal.

What is actually measured

This is the crux. Devices measure things that are not energy.

Step counts are measured reasonably well by accelerometers.

Heart rate is measured reasonably well at the wrist for steady activity, and less well during weight training or anything involving grip and wrist movement.

Energy expenditure is not measured at all. It is calculated from those inputs plus your entered age, sex, weight and height, using equations fitted to population averages.

Every source of individual variation — your actual efficiency of movement, your body composition, your fitness level — sits outside the equation.

How far off

Comparisons against laboratory measurement commonly find errors in the region of 20 to 30 per cent for energy expenditure across consumer devices, with individual sessions sometimes further out.

The direction matters as much as the magnitude: estimates tend to read high. A displayed 600 is frequently a real 450.

Why this specifically causes trouble

Because the number gets added to a food log.

Someone eating 1,800 kcal in a deliberate deficit sees 600 burned, eats back 600, and is now eating 2,400 against a real expenditure that included perhaps 450 of exercise. The deficit has shrunk by a third or vanished entirely.

Over weeks this looks like a metabolism that refuses to cooperate, and it is arithmetic. It is one of the reliable causes of the situation described in what a weight loss plateau usually turns out to be.

The strength training case is worse

For resistance training, the estimates are particularly poor. Wrist heart rate is unreliable during gripping and wrist flexion, the activity is intermittent, and the equations were built for steady-state aerobic work.

A watch reading 400 calories for an hour of lifting is offering a number with very little behind it. The actual figure for most people is considerably lower, and the training's value is not in the calories anyway.

What to do instead

Do not eat back exercise calories. Set an intake target from your own measured maintenance, which already includes the exercise you normally do. The method is in how to calculate your TDEE, and what it can't say.

If you must account for exercise, use half. Halving the displayed figure is a crude correction and closer to reality than accepting it.

Use the number relatively. Comparing today's session with last week's on the same device carries some information, because the systematic error is roughly constant. Comparing your watch's number with a friend's does not.

Track what is measured. Steps, duration, distance, heart rate, perceived effort. These are recorded rather than inferred, and they show progress, as covered in how to log cardio without lying to yourself.

The exception: relative daily expenditure

Wearables are more useful for spotting patterns across days than for pricing a session.

If your device shows you averaged 9,000 steps in March and 5,500 in April, that is a real and important change — and it is the kind of change that explains why a deficit stopped working, as described in steps, NEAT, and the calories you burn standing up.

Step counts are measured. Trust those more than the calorie figure derived from them.

The better instrument

Your weight trend measures the thing you actually care about, integrating every source of expenditure with no equation in between.

Two weeks of consistent intake plus a seven-day weight average tells you more about your real expenditure than any device will, and it costs nothing. That is the argument in daily weight lies, the trend does not.