Energy Balance

Energy balance is the arithmetic relationship between caloric intake and caloric expenditure across a time window. It underpins every nutrition recommendation for weight change: negative balance produces weight loss, positive balance produces weight gain, neutral balance maintains weight. This is thermodynamically guaranteed; where models and app predictions break down is in estimating the two sides accurately.

The three states

• Negative energy balance (deficit): intake < expenditure. Body draws on stored energy — primarily fat — to cover the shortfall.
• Neutral energy balance (maintenance): intake ≈ expenditure over time. Body weight stable to within normal water-shift fluctuation.
• Positive energy balance (surplus): intake > expenditure. Body stores the excess — as glycogen, as fat, and (with resistance training) as muscle.

Timescale matters

Energy balance is a function of the time window you integrate over. Hour-by-hour or meal-by-meal, most people oscillate between brief surpluses (post-meal) and brief deficits (overnight) even at steady-state maintenance. The meaningful question is: over a week, over a month, over a year, what is the cumulative balance?

Short-term tracking noise (water, glycogen, gut content) is irrelevant at these timescales. The long-run balance is what shapes body composition.

The "calories in" side

Intake is the more controllable of the two sides, but "more controllable" does not mean "easily measured":

• Food-database entries have ±5–10% typical variance against reference measurements.
• Manufacturer-labelled products have FDA-permitted ±20% tolerance.
• Portion estimation — eyeballed, photo-based, or even scale-logged — introduces additional error.
• Absorption varies: whole nuts, legumes, and some vegetables deliver fewer absorbed calories than their Atwater-factor label.

Realistic total error on a tracked daily intake: ±5% with meticulous kitchen-scale logging, ±10–15% with normal tracking, ±25–40% without tracking.

The "calories out" side

Expenditure is harder to control and harder to measure:

• BMR: predictive-equation estimate with ±5–10% individual error.
• NEAT: highly variable between individuals and within the same individual across deficit/surplus states.
• Exercise: wearable-estimated with ±10–25% error at individual level.
• TEF: typically embedded in the TDEE estimate implicitly.

Realistic total error on daily expenditure: ±10–20% from prediction alone, before behavioural-compensation effects.

The Pontzer "constrained total energy expenditure" model

Pontzer et al. (2016) observed that in a wide range of human populations — from sedentary office workers to Hadza hunter-gatherers walking 15 km/day — total daily energy expenditure varies far less than activity levels would predict. The explanation: the body appears to constrain total expenditure within a band, trading off active-activity energy against non-activity and maintenance energy. Heavy training does not linearly raise TDEE; the body partially compensates by reducing NEAT, immune activity, and reproductive-system energy costs.

Practical implication: "just exercise more" does not reliably deepen a deficit at the expected arithmetic rate. The compensation is real and substantial for many individuals.

Where this lands for trackers

The principle holds. The execution involves a lot of uncertainty on both sides of the equation. Empirical calibration — tracking intake against weekly-average-weight trend — is the way the uncertainty gets bounded.