4,200. That was the number my fitness tracker displayed at 10:47 PM last Tuesday. I had walked 8,300 steps, done a 45-minute ruck on the Butler Trail with my YVO Warrior group, and spent six hours at my desk in East Austin. The watch said I burned 4,200 calories. My spreadsheet said 2,890. The truth was somewhere in between, but closer to the spreadsheet. And after three months of daily comparisons, I stopped looking at the calorie burn number entirely. I do not trust it. I do not trust any calorie burn estimate from any consumer wearable. And here is the data that broke my faith.
I started this experiment in April because I was angry. My watch had been telling me I could eat 3,200 calories per day and maintain my weight. I was eating 2,800. I was gaining weight. The math did not work. Either physics was broken, or my watch was lying. I had a strong hypothesis about which one it was.
Here is what the data says. For 91 days, I wore three devices simultaneously: an Apple Watch Series 9, a Garmin Fenix 7, and a Whoop 4.0. I also tracked every calorie I consumed using a food scale and MyFitnessPal. I weighed myself every morning. I measured my BMR using indirect calorimetry at a sports lab in Austin (cost: $150, worth every penny). My measured BMR was 1,742 calories per day. My watch estimated it at 1,890. That is an 8.5% overestimation before I even got out of bed.
Calculate your daily calorie needs with TDEE and activity level adjustments.
All data stays in your browser — we never see it.The activity overestimation was worse. On a typical workday — 8,000 steps, no structured exercise — the Apple Watch estimated 680 calories from activity. The Garmin estimated 720. The Whoop estimated 590. My actual activity expenditure, calculated from heart rate reserve and movement data using the Compendium of Physical Activities, was 410 calories. The Apple Watch was overestimating by 66%. The Garmin by 75%. The Whoop by 44%. On days with structured exercise, the overestimation was even more dramatic. A 45-minute ruck with 20 pounds on the Butler Trail: Apple Watch said 612 calories. Garmin said 648. Whoop said 534. My actual expenditure, calculated from MET values and heart rate data, was 387 calories. The Apple Watch was 58% high. The Garmin was 67% high.
I was not surprised that the numbers were wrong. I was surprised by how consistently wrong they were. The Apple Watch overestimated every single day. The Garmin overestimated every single day. The Whoop overestimated on 87 of 91 days. This was not random error. This was systematic bias. The algorithms are calibrated to make you feel good about your activity. They are not calibrated to be accurate. They are calibrated to be encouraging. And encouragement is the enemy of precision.
The bias had a pattern. The overestimation was largest on low-intensity days and smallest on high-intensity days. When I did a hard interval workout, the Apple Watch was only 22% high. When I walked the dog around East Austin, it was 89% high. The algorithms seem to assume that any movement is more metabolically expensive than it actually is. They treat walking like jogging. They treat standing like walking. They treat fidgeting like exercise. And they add it all up into a number that makes you feel accomplished and hungry.
Calculate your Basal Metabolic Rate using Mifflin-St Jeor and Harris-Benedict equations.
All data stays in your browser — we never see it.And hunger is the real problem. If you believe you burned 4,200 calories, you eat 3,200 calories and feel virtuous. If you actually burned 2,890, you are in a 310-calorie surplus. Do that for a month, and you gain 2.6 pounds. Do that for three months, and you gain 7.8 pounds. I know because I did the math. And I know because I lived it. My weight crept up 4.2 pounds during the first six weeks of the experiment while I was "eating below my burn." I was not. I was eating above my actual burn by 200-400 calories per day. The watch made me feel like I was dieting. I was not.
I started using the calorie calculator on this site instead. I input my measured BMR, my activity level, my exercise frequency, and my goals. It gave me a TDEE of 2,650 calories for maintenance. When I ate 2,400, I lost weight. When I ate 2,800, I maintained. When I ate 3,000, I gained. The calculator was not magic. It was just not systematically biased toward making me feel accomplished. It was honest. And honesty is rare in the fitness tracking industry.
The heart rate data was actually quite good. All three devices were within 3% of chest-strap ECG readings during exercise. The step count was accurate within 5%. The sleep staging was mediocre but directionally correct. The calorie burn was the outlier. It was the one metric that was consistently, dramatically, and predictably wrong. And it is the one metric that drives eating behavior. People look at their calorie burn and decide what to eat. If the burn is wrong, the eating is wrong. The cascade is devastating.
Calculate your BMI instantly with visual category charts and history storage.
All data stays in your browser — we never see it.I brought this data to a meetup at Gevity on East Cesar Chavez. A woman named Rebecca, who works in UX design for a fitness app, said something that stuck with me: "We know the calorie numbers are wrong. But users get angry when we lower them. They feel like we are taking away their achievement." That is the tragedy. The devices are wrong because users want them to be wrong. We want to believe we burned 4,200 calories. We want to believe we can eat the burger. The market rewards optimism. The market punishes accuracy.
My ex-wife used to say I was no fun at parties because I would count the calories in the dip. She was not wrong. But I am not counting calories because I am obsessive. I am counting calories because I am honest. And honesty about energy balance is the only way to manage weight. The first law of thermodynamics does not care about your feelings. Calories in minus calories out equals weight change. If your out number is wrong, your entire equation collapses.
So here is my practical advice. Stop looking at your watch's calorie burn. Use a BMR calculator based on your actual measured data or a validated formula. Estimate your activity using MET values and honest time tracking. Use a food scale for everything you eat for two weeks to calibrate your eye. Then build a simple spreadsheet. TDEE = BMR + activity. Eat 500 below for loss. Eat at TDEE for maintenance. Eat 300 above for gain. It is not sexy. It does not give you a badge. But it works. And it is honest.
I still wear my watch. I still track my steps, my heart rate, my sleep, my HRV. Those metrics are useful. But the calorie burn number? I have hidden it. I have deleted the widget. I have trained myself to look away when the watch face shows it. Because 4,200 is a lie. And I am done believing lies, even when they come from a $400 device with a beautiful screen and a marketing team that knows exactly what I want to hear.
Pixel does not care about calorie burn. She cares about walks. And walks are good for you regardless of what number the watch assigns to them. That is the kind of metric I can trust. Simple. Honest. Measured in tail wags, not algorithms. I am thinking about building a dashboard that tracks tail wags per day. It might be more accurate than my Apple Watch.
