Math

How TDEE calibration works in CalBurndown

Every two weeks, we compare your predicted weight change to what actually happened, and adjust your TDEE accordingly. Here's how the loop works.

The Mifflin–St Jeor BMR equation is roughly accurate, but every individual is some percent above or below the prediction — typically ±10-15%. That gap is enough to make the difference between losing weight and stalling.

CalBurndown closes that gap with a biweekly calibration loop: we predict your weight change, compare to your actual, and adjust your TDEE multiplier.

The principle

If you've eaten X kcal over Y days and your weight changed by Z lb, we can solve for your actual TDEE:

TDEE = (X − Z × 3,500) / Y

Concrete example: you ate 28,000 kcal over 14 days, your weight dropped 1.5 lb.

TDEE = (28,000 − 1.5 × 3,500) / 14
     = (28,000 − 5,250) / 14
     = 22,750 / 14
     = 1,625 kcal/day

So your real TDEE for that window was 1,625 kcal/day. If your Mifflin-predicted TDEE was 1,750, your calibration factor is 1,625/1,750 = 0.93 — you burn 7% less than the formula says.

Going forward, we'll multiply your computed BMR by 0.93. Your burndown chart, your food budget, your buy-back calculations — all reflect your actual metabolism, not a textbook prediction.

Requirements before calibration runs

For the loop to produce useful numbers, we need:

  • At least 14 days of data since your last calibration
  • At least 50% of those days logged (food and activity)
  • At least one weigh-in in the window
  • Weight change between −10 lb and +10 lb (anything outside that range is almost always water/glycogen, not fat)

If any of those fail, we skip the calibration and try again next week.

What changes when calibration completes

The new factor is clamped to ±15% to prevent noise from breaking the math. So even if your data suggests a 30% adjustment (which is almost always a logging error, not real biology), we cap at ±15% and accumulate the rest over multiple cycles.

You'll see:

  • A small "TDEE updated" banner in the dashboard
  • The factor appears in Settings → TDEE calibration
  • Confidence label: low / medium / high based on data coverage

If you don't want calibration applied, Settings → TDEE calibration → "Use calibrated TDEE" can be toggled off. The factor still gets computed (so you can see what it would be), but your chart uses raw Mifflin–St Jeor instead.

Why two-week windows

Shorter windows (1 week) are too noisy — water weight, sodium, glycogen, and timing of the bathroom dominate the signal.

Longer windows (4+ weeks) catch genuine TDEE changes too slowly. If you start lifting heavier and adding muscle, your TDEE rises, but a 6-week window would lag badly.

Two weeks is the standard in metabolic research for catching real changes while filtering noise. We computed using a rolling window so calibrations don't have hard "reset" boundaries.

Common questions

My calibration says 0.87 — am I broken?

No. 13% below predicted means you have less lean mass than the formula assumes for your weight, or you sit more than you realize. Both are normal. The factor is what matters — apply it and your math becomes accurate.

My calibration says 1.14 — should I eat more?

Yes, if you're not losing weight as fast as you wanted. A 14% upward calibration means your real TDEE is higher than Mifflin predicts, often because you're more active (NEAT) than you realized. Eat to your calibrated TDEE minus deficit, not the textbook number.

How often does it run?

Every Monday at 5 a.m. UTC, but only for users who meet the data threshold. You can also force a recompute in Settings → TDEE calibration → "Recompute now."

Where to go next