November 11, 2019

The Science of Calorie Tracking

At WHOOP, we believe that you aren’t just an athlete during the couple of hours a day that you train or compete, but 24/7, and that part of being an Always On athlete means eating like one.

By Emily Capodilupo

In order to help guide our members towards appropriate fueling, we provide a daily estimate of calories burned. Recently, we made an improvement to this algorithm, and are excited to share the science and research behind how we count calories as well as take the opportunity to look at the challenges and limitations with calorie counting so that you can best understand how you should be actioning this metric.

Where Calories Come From and Where They Go

Calories are a unit of energy. We get our energy from food and expend that energy in three ways: (1) staying alive (2) digesting food and (3) doing things.An explanation of the 3 thrings that cause your body to burn calories.


The number of calories needed to maintain the basic functions of life – tasks like your heart beating, your eyes blinking, growing your hair and nails – is referred to as the “Basal Metabolic Rate,” or “BMR.” Your BMR depends in large part on your age and size. Women tend to have lower BMRs than men but this is mostly because they also tend to be smaller. Men and women of equivalent size and with equivalent lean muscle mass tend to have very similar BMRs. While you can increase your BMR by increasing your lean muscle mass, for the most part, this value is relatively consistent from month to month.


It turns out there is no such thing as a free lunch. If you want to digest your food, you will need to spend some calories in order to break it down into usable components. This physiological phenomenon is sometimes called the “Thermic Effect of Food” and collectively refers to the calories used to digest, absorb, and dispose of food. Some foods require more calories to process than others. For example, high fiber brown rice has a higher thermic effect than does its low fiber counterpart, white rice. Overall, about 10% of our total caloric intake is used to process food, but this varies significantly by diet, especially by the percent of protein in the diet. that 20-35% of the calories consumed as protein are used up digesting that protein.


Any level of activity beyond the basics of staying alive increases the rate of caloric consumption above the BMR. In general, the more you do, the more calories it takes to power that work. These calories are the ones you have the most control over and the primary reason why our caloric needs can vary so much from day to day. You can’t meaningfully change your BMR in a week, but add a 30-minute jog and you will increase your daily caloric consumption by anywhere from 250-550 calories.

Predicting Basal Metabolic Rate (BMR)

Over 100 years ago, Arthur Harris and Francis Benedict came up with the “” for predicting BMR from height, weight, age, and biological sex. While small changes to these formulas have been made in the century since, they still largely remain intact. However, their staying power should not be interpreted to be a sign of accuracy – the “Revised Harris-Benedict Equations,” which were and are considered the using only height, weight, age, and biological sex, is still known to have only a 95% confidence level of ±213 kcal/day for men and ±201 kcal/day for women.

Using incorrect biometric data, like an out-of-date weight, can further add to the error between your BMR estimate and truth. So in order to maximize accuracy, it’s important to make sure data like your height and weight are in your WHOOP app.

One common confusion is the difference between BMR and RMR (resting metabolic rate). BMR is, as mentioned above, the calories associated with maintaining the necessary functions of sustaining life (things like breathing) while RMR is used as an estimate of “ordinary” non-workout caloric consumption, and therefore includes all forms of “ordinary activity” (things like cooking breakfast). If you have ever used an online calculator that predicted your daily caloric needs and incorporated information like “How Active Are You” then that was an RMR calculator, not a BMR calculator. The practical difference is small, but in the WHOOP world, we use BMR and then account for things like the different caloric needs of someone with a desk job versus an Emergency Room nurse on their feet all day in our “Active Burn” calculation.

Predicting Active Burn

Estimating the caloric expenditure of activity is far trickier business than is estimating BMR, and even less relevant third-party research has been done in this area. The most accurate way to calculate calories burnt during exercise is through a process called “indirect calorimetry” which measures gas exchange (oxygen and carbon dioxide). Since indirect calorimetry requires use of a device covering the nose and mouth, such a technology is not available in wrist-worn devices. Working within the limitations of a wrist-worn device, calorie estimates using HR have been shown to far out-perform those using acceleration (motion) alone.

The most commonly used heart-rate-based formula for estimating caloric consumption during activity, or so called “active calories,” was . Despite being the best performing published algorithm of its kind, it has two primary issues – first it was fit only to exercise conducted at an intensity of 57%, 77% and 90% of Max HR and therefore untested above and below these intensities, and second it was found to explain only 73.4% of the variance in true energy expenditure. This means that a workout estimated to consume 1000 calories could have easily burnt 734 to 1266 calories – a massive difference if you are trying to use this information to modify your body composition. which did not include WHOOP confirmed this difficulty when it demonstrated that none of the commercially available wearables on the market at the time of the analysis could estimate workout calories within what the Stanford researchers had determined to be acceptable error.

Our active burn formula is inspired by the South African team’s methodologies and enhanced through our collection of millions of workouts worth of data across a wide spectrum of activities and intensities. In a recent update we modified our algorithm to better model the dynamics of the lowest-intensity exercises and to better integrate the active burn and BMR formulas by smoothing the transition between rest and exercise modes. This update primarily impacts periods of time in which your heart rate is between 30% and 40% of your heart rate reserve – the range from your resting heart rate (0%) to your max HR (100%) which is above resting/restful activities but below the level at which people typically exercise. This update has no impact on heart rates below 30% of your heart rate reserve and minimal impact on high-intensity exercise.

Predicting Caloric Consumption

You probably already know that if you consume more calories than you expend you will gain mass (either muscle or fat), if you consume fewer calories than you expend you will lose mass, and if you match input to output your weight stays the same. Because of this seemingly simple relationship, lots of athletes are interested in tracking their daily caloric expenditure in order to manage their weight according to their goals.

What you might not know is how imprecise commercially available estimates of caloric intake and expenditure actually are. The US Food and Drug Administration (FDA) requires that food labels are accurate within 20%, but that food labels are more likely to underestimate caloric content than they are to overestimate it. Calorie estimates on menus were shown to be of caloric inaccuracy, as changes in ingredient availability and cooking style lead to inconsistency in the preparation of even common dishes. The FDA’s seemingly loose 20% error bounds are also actually necessary as whole foods have natural variation in their caloric density – everything from weather, soil quality, health of the plant, and ripeness contributes to how nutrient and calorically dense produce is. Therefore, not surprisingly, natural foods – like a salad – tend to differ from their advertised caloric contents by more than processed foods – like a candy bar – which tend to be truer to their nutrition labels.

What this all means is that, even if you weigh every gram of food that you put in your body, you couldn’t possibly count your caloric intake with smaller than a 20% error bound. Those 2000 calories you believe you consumed could easily have been 2400, an error bar approximately the size of a .

So Where Does This Leave Us?

Regardless of what any calorie tracker is advertising, estimates of caloric consumption need to be understood to be just that, estimates. But, that doesn’t mean they aren’t useful. Heart-rate-based calorie estimates, like the ones provided by WHOOP, scale reliably with true caloric consumption, meaning that you can look at trends in caloric burn and trust that it’s accurately identifying relatively high and low calorie-burn periods. This reliability is key to its use, because it means that eating more or eating less on days when you burn more or fewer calories is still a reasonable way to action this data.

On the ingestion side, it is important to understand all the limitations on accurate calorie counting, even with the most diligent and intentional tracking practices:

  • Nutrition labels are estimates, not guarantees, and in fact are and very rarely audited for compliance with this lax standard.
  • Calorie estimates from restaurants are statistically more likely to underestimate total caloric content than to overestimate them.
  • Caloric estimation guidelines for produce and whole foods are often decades old and may not reflect changes in nutritional profiles caused by industrial scale agriculture and modern farming practices.
  • The way we prepare food can alter its caloric density. For example, that we derive more calories from cooked meat than we do from raw meat.
  • Difficult (if not impossible) to measure factors including the – that is the ecosystem of bacteria that live in your digestive tract – influence the proportion of food calories that actually get absorbed.

A more exhaustive review of all challenges with accurately counting calories can be found in a .

For all these reasons, WHOOP always recommends mindful eating – that is paying attention to portion sizes and checking in with yourself if you feel full.

Share on and

Emily Capodilupo

Before joining WHOOP in 2013 as the first full-time employee and first scientist, Emily studied Neurobiology at Harvard University and studied circadian biology in the Analytical and Modeling Unit of the Division of Sleep Medicine at Harvard's Brigham and Women's hospital. As a runner and acrobat, Emily knows first hand the importance of sleep and recovery for peak performance. At WHOOP, she blends this personal experience with the sleep and analytics knowledge developed at Harvard to empower athletes to make intelligent, data-driven decisions.