Where the numbers come from

Last updated 23 September 2026

Roughly estimates, and it says so. But an estimate still has to come from somewhere. Every figure the app puts in front of you is below, with how it is worked out and the published work it rests on. The same list is in the app, under Settings → Where the numbers come from.

Your starting daily calorie target

Estimated with the Mifflin-St Jeor equation from your height and weight, then multiplied by 1.4 — the low end of the FAO/WHO/UNU "sedentary or light activity" band — because someone who has not set a target has not told us how they train. The equation needs age and sex, which the app never asks for: it assumes 35 and the midpoint of the male and female constants, so the answer is wrong by a similar margin in either direction instead of being right for one and badly off for the other. It is a starting point to adjust, not a prescription.

Your starting daily protein target

Set at 0.8 g per pound of bodyweight, which is about 1.76 g/kg. Morton and colleagues pooled 49 trials and found gains from added protein plateaued around 1.62 g/kg/day, with the confidence interval reaching 2.2 — so this sits just inside the useful range rather than at the bottom of it. The ISSN position stand recommends 1.4–2.0 g/kg for anyone training.

How processed the day’s food was

The gauge places each food on the NOVA classification — 1 unprocessed, 2 culinary ingredients, 3 processed, 4 ultra-processed — and averages them weighted by calories. NOVA is the scale used in nutrition research, not something invented here. It is a description, not a verdict: protein powder and supermarket bread are both group 4 and both fine.

Fat quality — “whole food” against “refined”

This axis is the food the fat came in, not how saturated it was. Saturation puts crisps and avocado in the same place, because crisps are fried in sunflower and corn oil and are about 85% unsaturated, and makes butter the worst thing on the plate. The evidence has moved towards judging the whole food rather than the fatty acid in it, which is the distinction people actually mean.

Caffeine still active, and when it fades

Caffeine clears by first-order elimination — a fixed fraction per unit time — so each dose is decayed by its own age and the results added. The app uses a five-hour half-life, the usual adult mean; the real range is roughly 1.5 to 9.5 hours and moves with pregnancy, oral contraceptives, liver function and smoking, so treat the figure as a guide to your own pattern rather than a measurement. The FDA puts 400 mg a day as the amount not generally associated with harm in healthy adults.

Calories and protein read from a photo or a sentence

An estimate, and the app is named after how good it is. Energy is converted with the standard Atwater factors — 4 kcal per gram of protein and carbohydrate, 9 for fat — applied to a portion judged from a picture or from what you typed. Even a printed label is a tolerance rather than a measurement: US regulation allows the calories in a packaged food to run up to 20% above the number on the box.

This is not medical advice

Roughly is a general-wellness coach, not a clinician. It does not diagnose, treat, prevent or monitor disease, and nothing it says is medical advice. Its figures are estimates from what you tell it, never measurements. Talk to a doctor or a registered dietitian before changing how you eat or train — especially if you have a health condition, are pregnant, take medication that food or exercise affects, or have any history of disordered eating.

Think a number is wrong?

Say so — a citation is only useful if it can be argued with. myhumblecoder@gmail.com.