Energy balance, written as arithmetic
Over any stretch of weeks, the energy you took in either got burned or got stored. Rearranged, that gives the estimate:
energy burned per day ≈ average intake − (energy stored or released ÷ days)
Here’s a worked case. Across 21 days with food logged in full, intake averages 2,100 kcal a day, and the weight trend drops 1.5 lb over the same three weeks. Using the common rough figure of 3,500 kcal for each pound of loss, that drop released about 5,250 kcal, or 250 kcal a day. Expenditure comes out near 2,350 kcal a day.
The 3,500 figure is an approximation. Body fat stores about 39.5 MJ per kilogram (roughly 4,300 kcal per pound) and lean tissue 7.6 MJ/kg (around 800 kcal per pound), so the true value depends on what mix of tissue changed (Hall et al., 2011). The same paper warns against using the 3,500 rule to forecast long-term loss, because it ignores how the body adapts as weight falls.
Why the trend goes into the equation
Feed raw scale readings into that arithmetic and the answer jumps around. A 2 lb difference from water at either end of a three-week window shifts the result by more than 300 kcal a day. Each gram of stored glycogen carries several grams of water with it, and the first days of a sharp calorie cut often draw that store down, so early weigh-ins can make a deficit look larger than it is (Kreitzman, 1992).
The weight trend separates those swings from the underlying direction, which is the quantity the equation needs. Its weekly rate comes with a range, and that range flows straight into the energy estimate: an uncertainty of 0.25 lb/wk in the rate works out to about 125 kcal a day.
Formulas are a starting guess
Before any of your own data exists, an estimate has to come from somewhere. Most calculators start with a resting metabolic rate equation such as Mifflin-St Jeor, then multiply by an activity factor. A systematic review found Mifflin-St Jeor the most reliable of the common equations, landing within 10% of measured resting rate for more people than the others, while still noting meaningful errors for individuals and thin validation in older adults and some ethnic groups (Frankenfield et al., 2005).
The activity multiplier adds a second guess on top of the first. A formula is a reasonable place to begin, and your own intake and trend can replace it once they say more about you than a population average does.
A missing food day can’t count as zero
Now suppose 3 of the 21 days had gone unlogged. Count them as zero and the average intake falls from 2,100 to 1,800 kcal a day, so the expenditure estimate drops by 300 kcal along with it. Any calorie target built on that number would then ask for less food than it takes to hit the chosen rate.
Self-reported intake also tends to run low even on days that feel complete. In an NIH study using biomarkers, 24-hour recalls underreported energy by about 12–14% for men and 16–20% for women (Subar et al., 2003). An adaptive estimate compares your logs with your own trend, so a steady under-logging habit lowers the estimate by a similar share. The result stays usable for setting intake as long as the habit stays consistent.
How many complete days it takes
Expect weeks. The energy estimate can only be as precise as the rate of the trend, and that rate needs a run of weigh-ins before its range tightens. The early water movement from glycogen also argues for letting the first week or so wash through before reading much into the number.
Sparse weigh-ins lengthen the wait, and so do frequent incomplete days. Expenditure also drifts as weight changes, since energy needs drop alongside body weight (Rosenbaum and Leibel, 2010), so a good estimate keeps updating as the months go by. The plateau guide explains how that drift bends a long campaign.
How Balance Point estimates energy burned
In Balance Point, each day’s intake is four totals (calories plus the three macronutrients), entered by hand or imported from Apple Health. Days carry a complete or incomplete flag, and an unlogged day never enters the average. The energy estimate opens with a formula, then switches to the intake-and-trend method described above once there are enough complete days to support it.
Its status reads formula, adapting, stable, or paused, and a confidence figure sits beside the number. Intake targets cover holding weight, losing, and gaining, with guardrails on each and a seven-day check-in cycle. Balance Point doesn’t include a food database or log individual meals, so the daily totals are the whole input. The energy feature has the details.