Your Watch Says You Burned 620 Calories. The Research Says Assume Less.

You finish a 45-minute session and the watch reports 620 calories. That's a large meal's worth of permission, and if you're in a deficit it's tempting to treat it as earned.

Here's the problem: heart rate is the only physiological variable most wrist devices actually measure, and heart rate is a poor proxy for energy expenditure. Everything else in that 620 is inference from population equations applied to your height, weight, age and sex. Validation studies comparing consumer wearables against indirect calorimetry — the laboratory reference — have repeatedly found step counting and heart rate to be reasonably accurate while energy expenditure estimates carry error rates that are far larger and skew toward over-reporting.

The number isn't useless. It's just not the number people think it is.

Why Energy Expenditure Is So Hard to Estimate

Heart rate and calorie burn aren't the same signal. Heart rate rises with heat, dehydration, caffeine, stress, poor sleep and psychological arousal — none of which burn meaningful energy. A stressful meeting can put your heart rate where a brisk walk would, at a fraction of the energy cost. The device sees the beats, not the reason.

Optical sensors have their own error. Wrist-based photoplethysmography works by reading light reflected from blood flow. It performs reasonably during steady-state effort and degrades during intervals, rapid transitions, and any activity involving wrist movement or grip — weightlifting, rowing, boxing. Skin tone, tattoos, fit and ambient temperature all affect the signal. A chest strap removes most of this specific error source, though not the estimation problem behind it.

Individual efficiency varies substantially. Two people of identical height, weight, age and sex can differ meaningfully in the energy cost of the same task, driven by gait mechanics, training status and body composition. Trained runners are more economical than untrained ones — they burn less for the same pace, which is the whole point of training and precisely what a population equation cannot know about you.

Resistance training is nearly unmeasurable this way. Lifting involves short bursts, long rests, high perceived effort and elevated heart rate driven partly by pressure response rather than aerobic demand. Wearable estimates for strength sessions are the least reliable of all, frequently by a wide margin.

The Error That Matters Most

Even a device with modest percentage error causes a specific and common problem, because of what people do with the number.

Total burn includes what you'd have burned anyway. If your watch reports 620 calories for an hour of exercise, some of that is your resting metabolic rate — energy you'd have spent sitting on the sofa. For a 75 kg adult, resting expenditure is roughly 70–80 calories an hour. The additional cost of the exercise is 620 minus that: closer to 545. Some apps report net, most report gross, and few make it obvious which. This alone accounts for a systematic overestimate of what a session "earned."

Eating back exercise calories compounds every error. The classic failure: TDEE estimated slightly high, exercise burn estimated 25% high, food intake under-reported by 20% — which research on self-reported dietary intake finds to be routine and largely unintentional. Each error is individually modest. Together they can erase a 500-calorie deficit entirely, and the person doing everything right sees no results and concludes their metabolism is broken.

A worked example. Target deficit 500/day. Watch reports 620 burned; true additional cost is perhaps 450. You eat back 620. Net effect: 170 calories over. Add typical under-reporting of intake and the deficit is gone.

Getting a Number You Can Use

Start with MET values rather than a wrist estimate. The Compendium of Physical Activities assigns each activity a metabolic equivalent — its energy cost relative to rest. Calories per minute ≈ MET × 3.5 × body weight in kg ÷ 200. Walking at 3.5 mph is about 4.3 METs; running at 6 mph is about 9.8; general weight training is about 3.5–5. For a 75 kg person running 30 minutes at 6 METs: 6 × 3.5 × 75 ÷ 200 = 7.9 calories/minute, or about 236 calories. The calories burned calculator applies this properly for common activities.

MET values are population averages too, but they're transparent averages — you can see the assumption, which you can't with a proprietary wrist algorithm.

Then discount what you were going to burn anyway. Subtract roughly 1.0 MET × the duration to convert gross to net. It's a small correction that removes a consistent bias.

Use the trend, not the reading. Your device's absolute number may be off, but its error is broadly consistent for the same activity on the same body. If it said 600 last month and says 660 for the same session now, you probably are working harder. Devices are far better at relative comparison than absolute measurement.

What to Actually Do

Set your target from TDEE, and don't add exercise on top. Calculate maintenance with the TDEE calculator using an activity multiplier that already reflects how much you train. Set your deficit from that single number and eat the same amount every day, training or not. This removes the entire eating-back problem, and it's easier to follow than a moving daily target.

If you must eat back, take half. For those who genuinely need the fuel — endurance athletes, high-volume trainees — eating back 50% of the reported burn is a reasonable hedge against the systematic overestimate. It's arbitrary, but it's arbitrary in the right direction.

Let the scale correct the estimate. All of this is theory until it meets your actual results. Hold intake steady for three weeks, track weekly average weight, and adjust from what happens. If the average is flat when you expected 0.5 kg a week, your true intake or expenditure is different from the model — by exactly the amount you observe. That measurement beats every equation, including the ones in this article.

Use wearables for what they're genuinely good at. Steps, heart rate zones during steady cardio, sleep duration, resting heart rate trends, training consistency. All of these are useful and reasonably measured. Calorie burn is the single weakest output on the device, and it's the one displayed most prominently.

The watch is a decent training instrument and a poor accounting instrument. Set your intake from a maintenance estimate, let the scale audit it over weeks, and treat the 620 as motivation rather than currency.