Beyond population averages

Adaptive TDEE Calculator

Standard calculators predict a population average, then assume your metabolism is static. This one prioritises lean body mass, reads your week in plain English, and adjusts for NEAT, plasma donation, and metabolic adaptation. Every formula is tied to the sources cited below.

Your numbers

Leave blank to use Mifflin-St Jeor. Enter a value to switch to the lean-mass Katch-McArdle equation — more accurate for lifters and anyone whose body composition is far from average.

Typing is parsed instantly by a keyword engine. For messy or unconventional schedules (“two-a-days Tue/Thu, coach soccer on weekends”), click Understand with AI to send the text to Claude for a smarter read. Either way the calorie math is done by the same tested engine — the model only extracts the structure, which is shown below so you can correct it.

Override the base activity multiplier (advanced)

Prolonged deficits trigger adaptive thermogenesis — expenditure drops beyond what tissue loss predicts and lingers afterward (Trexler 2014). A surplus nudges it slightly up. Applied as a modest multiplier.

Project a weight goal (optional)

Diet & macro planner

Slide to pick your intensity — from a 1,000 kcal cut to a 1,000 kcal bulk. Macros are built from your exact maintenance ( kcal), with protein set per Stronger By Science (scaled to your lean mass when body-fat % is entered).

−1000 · cut maintenance +1000 · bulk
Maintenance kcal/day

Protein — Stronger By Science (Nuckols); corroborated by Roberts/Helms 2020 (1.8–2.7 g/kg, up to 3.5). Fat — 15–25% of calories (Helms 2020), floored at 0.5 g/kg for hormones. Carbs — the remainder; resistance trainees don't need endurance-style carb loads (Escobar 2017).

Research & pitfalls: why most TDEE numbers are wrong

A TDEE estimate is only as honest as its assumptions. Below is what the literature actually says about where these calculators break down — and how this tool tries to do better. Citations refer to the sources listed at the end.

1. They predict a population, not you

Equations like Mifflin-St Jeor and Harris-Benedict are regressions fit to groups. They work “reasonably well across large groups, but an individual can easily be 10–20% above or below the prediction” [Problems]. A true maintenance of 2,800 kcal could be predicted anywhere from ~2,400 to ~3,200. The Pontzer life-course dataset is blunter still: total expenditure varies more than ±20% between people even after controlling for fat-free mass, fat mass, sex and age [Pontzer 2021]. That irreducible spread is why we show a range, not just a point.

2. Activity buttons are too blunt

“Sedentary / lightly / moderately / very active” are vague. A student who lifts five days a week but sits otherwise may pick “very active,” while a warehouse worker with no gym time picks “moderately active” and is badly under-counted [Problems]. We instead build a Physical Activity Level from your described job (PAL bands from the FAO/Black tables: seated 1.4–1.5, standing work 1.8–1.9, strenuous 2.0–2.4 [Estimating]) and add each workout separately via METs [Computational].

3. Exercise calories are inflated

Trackers, cardio machines and online calculators routinely overstate burn: “someone who believes they burn 800 calories during a workout may actually burn 400–600” [Problems]. Two guards keep us honest — we use moderate MET values, and we count only the net cost above rest (MET − 1), because the resting calories during that hour are already inside your base multiplier.

4. NEAT is the hidden hundreds

Non-Exercise Activity Thermogenesis — walking, fidgeting, standing, chores — is the single largest swing factor. “Two people with identical workouts can differ by hundreds of calories per day because one naturally moves much more than the other” [Problems]. In sedentary people EAT can be ~0% of expenditure; NEAT makes up the remainder [Estimating]. That’s why a step count moves your number here, and why the honest range stays wide.

5. Total weight ≠ metabolic tissue

Fat-free mass is what correlates with resting metabolism, “independent of age, BMI, and other metabolic variables” [One2One]. Two 158-lb people at 33% vs 45% body fat had RMRs of 1,571 vs 1,252 kcal/day [One2One]. Equations on total weight alone can’t see that. With a body-fat input we switch to Katch-McArdle (370 + 21.6 × lean mass) and cross-check it against the FFM equation 23.9 × FFM + 372 [Estimating] and the Pontzer FFM power law [Pontzer 2021].

6. Metabolism isn’t static

After a prolonged deficit, every component of expenditure falls — BMR, NEAT, the thermic effect of food, even the cost of moving — by more than lost tissue predicts. This adaptive thermogenesis persists after weight stabilises, and its size is proportional to the deficit [Trexler 2014]; very-low-calorie dieting slows metabolism to protect vital organs [One2One]. Overfeeding pushes the other way (~+7% acutely [Trexler 2014]). The diet-history selector applies a modest version of this.

7. On plasma donation — where the numbers actually come from

This figure is derived from biochemistry, not invented. Plasma is 91–92% water and 8–9% solids, carrying ~70–90 g/L of protein [StatPearls][Deranged Physiology]. A plasma session keeps ~690–880 mL (more for heavier donors), removing roughly 50–66 g of protein. The cost has two parts: the caloric content that leaves (~4 kcal/g) and the energy to resynthesise it (~2.2 kcal/g). A PhD-biochemistry itemisation pegs a whole-blood unit at ~425–460 kcal of content, “higher” once synthesis inefficiency is added [Maynard]; an apheresis director cites ~500 kcal to replace a donation [Columbia] and UC San Diego ~650 [UCSD]. Those higher figures are for whole blood, most of which is hemoglobin. Plasma donation returns your red cells [StatPearls], so the plasma-specific number is lower, ~355–455 kcal — which is what this tool computes, scaled by your weight. It remains the most variable input (hydration, diet, protein level, donation type), so it’s the one we let you tune directly.

Finally, the honest disclaimer no parser can remove: self-reported activity is biased. People over-report exercise and under-report sitting; structured-text parsing makes the estimate granular and transparent, but it cannot make a rough self-description accurate. Treat the output as a hypothesis to test against two weeks of weigh-ins, not a measurement. The gold standard remains indirect calorimetry or doubly-labelled water [Estimating].

8. BMI is a population tool, not a verdict on you

The BMI shown above is just weight ÷ height² — a fine screen across large groups, but it cannot tell muscle from fat. A lean 180-lb lifter and a sedentary 180-lb person get the identical BMI, and muscular athletes are routinely flagged “overweight” or even “obese.” It also ignores fat distribution, age, sex, and ethnicity. This is the same lesson as the rest of the tool: total weight hides body composition. Read your BMI as rough context, and trust your body-fat % and the mirror over the category label.

9. How the macro targets are built (and why the deep cut is red)

Protein comes first, set per Stronger By Science [SBS]: ~2.0 g/kg bodyweight to maintain or gain (women ~1.75), and — when you enter body-fat % — scaled to fat-free mass (~2.35 g/kg FFM), which SBS argues is the more accurate anchor. Dieting raises the target (≥2.5 g/kg FFM, up to 4.0) to defend lean mass, and the benefit is larger the leaner you are. This converges with Roberts/Helms’ 1.8–2.7 g/kg (up to 3.5) [Roberts 2020]. Fat is set at 15–25% of calories (a little lower on a cut) with a 0.5 g/kg floor for hormones [Roberts 2020], and carbohydrate takes the remaining “energy budget.” Resistance trainees genuinely don’t need the 4–7 g/kg carb loads borrowed from endurance sport — low-carb doesn’t impair lifting performance or hypertrophy signalling [Carb Need].

The slider turns red toward the −1,000 deficit for a reason: a gradual loss of 0.5–1% of bodyweight per week best preserves muscle [Bulk/Cut], and the size of the deficit drives the size of the metabolic adaptation and muscle loss [Trexler 2014]. The badge tells you your personal weekly rate, since “too aggressive” depends on your bodyweight — a −1,000 deficit is ~1.1%/week for an 80 kg lifter (danger) but only ~0.9% for someone at 100 kg. These macros are a research-grounded starting point; adherence and two weeks of real-world adjustment matter more than the decimals.

Sources

  1. [Problems] Problems with Modern TDEE. Provided briefing on population-average error, activity-category vagueness, exercise overestimation, NEAT, body composition, and metabolic adaptation.
  2. [Estimating] Estimating Energy Needs for Research Diets. Recommends Mifflin-St Jeor over Harris-Benedict; FFM equation REE = 23.9·FFM + 372; PAL tables (Black et al.; FAO/WHO/UNU).
  3. [One2One] Resting Energy Expenditure (ESSCO2O / one2one). REE = 60–70% of TEE; FFM drives REE; REE declines ~2–3%/decade after 20; body-composition RMR example.
  4. [Pontzer 2021] Pontzer H. et al. Daily energy expenditure through the human life course. Science 373(6556):808–812. TEE = 0.677·FFM0.708 (r²=0.83); ±20%+ residual variation; decline after ~60.
  5. [Trexler 2014] Trexler E.T., Smith-Ryan A.E., Norton L.E. Metabolic adaptation to weight loss: implications for the athlete. J Int Soc Sports Nutr 11:7. Adaptive thermogenesis across BMR/NEAT/TEF/EAT; persistence; ~+7% from refeeding.
  6. [Computational] Computational Methods with MET. Mifflin worked example; MET kcal/min = MET·kg·3.5/200; body-mass energy density 7700 kcal/kg.
  7. [StatPearls] Mathew J, Sankar P, Varacallo M. Physiology, Blood Plasma. StatPearls, 2023. Plasma = 91–92% water, 8–9% solids; in plasmapheresis the red cells are returned to the donor.
  8. [Deranged Physiology] Yartsev A. Constituents and functions of plasma. CICM Primary. Plasma protein ~70–90 g/L (~80% albumin).
  9. [Maynard] Mark DA. Calories in Human Blood. Maynard Life Outdoors, 2010. Itemised ~425 (women)/460 (men) kcal of content per 500 mL whole-blood unit; replacement cost higher due to synthesis inefficiency.
  10. [Columbia] The Surprising Benefits of Donating Blood. Columbia University Irving Medical Center / NewYork-Presbyterian, 2022. Apheresis director: “it takes your body about 500 calories to replace” a donation.
  11. [UCSD] Why Donating Blood Is Good For Your Health. Medical Daily, 2013, citing UC San Diego: “~650 calories per pint.”
  12. [SBS] Nuckols G. Protein Science Updated: Why It’s Time to Move Beyond the “1.6–2.2 g/kg” Rule. Stronger By Science. ~2.0 g/kg BW (men) / 1.75 (women) or ~2.35 g/kg FFM to maximise gains; dieting ≥2.0 g/kg BW / 2.5 g/kg FFM, up to 3.0 / 4.0.
  13. [Roberts 2020] Roberts B.M., Helms E.R., Trexler E.T., Fitschen P.J. Nutritional Recommendations for Physique Athletes. J Hum Kinet 71:79–108. Protein 1.8–2.7 g/kg (up to 3.5); fat 15–25% of calories; carbs 2–5 g/kg from the remaining energy budget.
  14. [Carb Need] Escobar K.A., VanDusseldorp T.A., Kerksick C.M. Carbohydrate intake and resistance-based exercise: are current recommendations reflective of actual need? Br J Nutr 116:2053–2065. Resistance training doesn’t deplete glycogen like endurance; lower carb intakes don’t impair performance or hypertrophy signalling.
  15. [Bulk/Cut] Preiato D. (rev. Tinsley G.). Bulking vs. Cutting. Healthline, 2021. Bulk +10–20% (0.25–0.5%/wk gain); cut ~500 kcal deficit, gradual 0.5–1%/wk loss best preserves muscle; protein 1.4–3.1 g/kg.
  16. [Nutrient Timing] Kerksick C.M. et al. ISSN position stand: nutrient timing. J Int Soc Sports Nutr. Daily protein/carbohydrate totals dominate; distribute protein across the day around training.

This tool is for education and self-experimentation, not medical advice. Energy needs in pregnancy, growth, illness, or endocrine disorders fall outside these equations — consult a qualified professional.