Methods and choices
Ballpark generates calorie targets, protein targets, and ranges. This page says where each one comes from. Three labels run down the page, and every claim carries one of them.
- Published A named study or reference standard says this. The source is linked.
- Measured Ballpark's own data says this. The test that produced it is described.
- Our choice The research doesn't settle this, so we picked something. You get to see what and why.
Logging a meal
A guess never gets promoted to a fact.
A barcode, a nutrition label, or a weighed amount is exact and stays exact. A photo or a description becomes an estimate with a likely range attached, and you can correct the food, the amount, or both before saving. When a day mixes the two, the day keeps a range rather than collapsing into one confident total. A day you didn't finish logging is marked incomplete rather than counted as zero.
Everything below rests on that. A number's evidence travels with it, so the app can always tell you how much weight it deserves.
Your weight trend
The equations get you started. Your own weight takes it from there.
Published Even with perfect inputs, a population equation can miss an individual by hundreds of calories a day.
The 2023 National Academies report on energy needs publishes the spread around its own predictions, and that spread is wide. It isn't a flaw in the equation. It's what happens when you predict one person from a population average.
Our choice The starting number is a hypothesis, and your weight is the test.
Every calculation further down this page produces a starting point and nothing more. Once you have a few weeks of weigh-ins, Ballpark compares the direction you asked for against the direction you're actually going, and proposes a new target from that. Your weight already accounts for everything the equation guessed at, including the meals you never logged.
This is why the sections below matter less than they look. They decide where you start, and your own data decides where you end up.
When Ballpark won't estimate
Ballpark stops instead of guessing.
Both equations need a sex input. You can decline to give one, and some people will want to.
Measured Without it, the answer spans about a thousand calories a day.
For a representative adult, combining both equations produces a range of 2,122 to 3,076 calories. A target anywhere in that band would look just as confident as any other number on the screen.
Our choice Choosing "prefer not to say" sends you to enter your own target.
Ballpark could pick one equation without telling you, or average the two into a number with no research behind it. It does neither. It tells you the estimate can't be made and hands you the control. This is the clearest case on the page of a citation deciding how the app behaves.
No exercise calories added back
Published No wrist-worn device tested in a Stanford study came within 20% on calories burned.
A 2017 Stanford validation study measured seven consumer wearables against clinical gold standards across sixty people. Six of the seven tracked heart rate within 5%. Not one estimated energy expenditure within 20%, and the authors advised caution about using those numbers in health programs. Heart rate is measured. Calories burned are modeled from it, and that is where the error enters.
Our choice Wearable burn estimates are too noisy to spend against a target.
Adding a watch's calorie estimate to your daily allowance gives you a larger number to eat toward, built on a measurement with a wide error bar. Your expenditure is already reflected in what the scale does over the following weeks, which is the signal Ballpark actually uses.
Your starting calorie target
Two equations, averaged.
Published Your starting calorie estimate is built from two sources.
- The 2023 National Academies report on energy needs. It's the current U.S. reference for how many calories people need, and it's built from studies that measured what people actually burned, using a tracer method called doubly labeled water, rather than predicting it from height and weight alone.
- Mifflin-St Jeor. A 1990 equation for resting energy, multiplied by a conventional activity factor. It's the most widely used clinical estimate and the number most other calculators are built on.
Our choice Ballpark averages them, weighted equally.
The National Academies equations ran high against the maintenance histories we could check, and resting energy times an activity factor tends to run low. Averaging them starts closer for more people. Both original numbers are stored with your profile, so the average never becomes the only surviving record.
The average is a product decision rather than a third equation, and it has no published validation of its own. It stays a starting point until your own weight data has something to say.
Measured Averaging also fixed a bug we could see in the raw equations.
We ran both equations across every adult body: ages 19 to 100, heights 145 to 205 cm, weights 40 to 200 kg, both equation sexes, and every activity category. The National Academies equations alone produced 574 cases where moving from "active" to "very active" predicted fewer calories, which is backwards. The largest was 77 calories a day.
The average has none, so Ballpark doesn't need a correction layer on top. That sweep runs as a test on every build.
Our choice Generated targets round to the nearest 50 calories, and protein to the nearest 5 grams.
A target of "2,663 calories" claims a precision no equation here has. The unrounded number stays in the calculation record and never reaches the screen.
Why the target is a range
One number would be hiding something.
That average is one number, and a single number implies the equations know more about you than they do.
Published Two people with identical height, weight, age, and activity can differ by hundreds of calories a day.
The National Academies report publishes how far off its own predictions typically land for an individual. That spread is real, it's large, and most calculators drop it before showing you anything.
Our choice Ballpark keeps that spread and centers it on the average.
Continuing the example above, the published error spread around an average of 2,722 puts the honest range at 2,380 to 3,064.
Ballpark calls this a rough range and never attaches a confidence percentage to it. The spread was fitted to the National Academies equation, not to our average of two equations, so quoting it as a statistical interval would be borrowing rigor that doesn't transfer. If a wider set of real maintenance data ever supports a better width, this gets replaced.
From pace to calories
Choosing a pace shifts the starting number, and does nothing else.
Published The familiar conversion is about 7,700 calories per kilogram, or roughly 3,500 per pound.
Split across a week, that comes to about 250 calories a day for half a pound a week, and about 500 for a pound.
Published That same conversion is wrong as a prediction.
Hall and Chow's 2013 paper is titled "Why is the 3500 kcal per pound weight loss rule wrong?" and the answer is that your expenditure changes as your weight and intake change. The rule is a reasonable way to pick a starting number and a bad way to forecast a date.
Our choice Ballpark never shows a completion date or a future weight.
There's no countdown, no arrival estimate, and no curve sloping toward a goal weight. Those are the easiest things in this category to build and the least defensible.
Suggested loss and gain rates
Both numbers come from the same question: how fast can body weight change before the change stops being mostly fat?
Our choice Pace is reasoned in percent of body weight, not pounds.
Half a pound a week is 0.50% of body weight for someone at 100 pounds and 0.12% for someone at 400. Advice given in fixed pounds stops applying as body size changes, so both thresholds below are percentages, and Ballpark displays whichever unit you use.
Losing: up to 1% of body weight per week
Published Losing at half the speed protected lean mass in a randomized trial.
Garthe and colleagues randomized 24 elite athletes in 2011 to lose weight at 0.7% or 1.4% of body weight per week, with four resistance sessions a week in both groups. Both groups lost about the same total weight, near 5.5%. Lean mass rose 2.1% in the slower group and was flat in the faster one, a difference the authors report at p < .01.
Published Reviews land on 0.5 to 1% a week.
A 2021 narrative review by Ruiz-Castellano and colleagues sets calories against a target of 0.5 to 1.0% body weight per week to maximize fat-free mass retention, and adds that leaner people should take the smaller deficit. The 2017 ISSN position stand reaches the same place from the same trial.
Our choice The note appears above 1% a week, or above two pounds.
Ballpark uses the top of the published range rather than the middle, because a slower loss is a preference and not a correctness rule. The two-pound check catches a pace that looks gentle as a percentage but is large in absolute terms. As with gaining, the note is dismissible and the pace stays yours.
Gaining: 0.25 to 0.5% of body weight per week
Published This range comes from a 2019 review of off-season bodybuilding nutrition.
Iraki and colleagues recommend a 10 to 20% energy surplus and 0.25 to 0.5% of body weight per week, with more experienced lifters at the lower end because there's less muscle left to add. It's a narrative review, and the authors say directly that the underlying literature is thin.
Published The evidence is clearest about what happens when you gain faster than this.
Helms and colleagues ran 21 trained lifters through eight weeks of supervised training in 2023. Muscle thickness and squat strength came out similar across surplus sizes, while faster weight gain strongly predicted thicker skinfolds, meaning fat. Garthe and colleagues found the same shape in elite athletes in 2013: the guided group ate about 621 more calories a day and gained 3.9% body weight against 1.5%, with fat mass up 15% against 3% and no significant difference in lean mass.
Published Nobody has validated an optimal surplus.
Slater and colleagues concluded in 2019 that the commonly cited "sweet spot" has never been tested in trained lifters. A 2021 Bayesian analysis by Smith and colleagues points the other way from our default, putting entirely lean gain at around 0.55% a week, though the authors call it preliminary and the interval was wide enough to include both fat loss and fat gain.
Our choice The note appears above 0.5% a week.
That is the top of the range Iraki recommends, so the note marks the edge of the published advice rather than a verdict on your plan.
The protein target
0.8 g per pound, or 0.9 when you're losing.
Published These numbers were established in people who lift.
Morton and colleagues' 2018 meta-analysis measures what protein does for resistance-training adaptations, and Helms and colleagues' 2014 review covers trained lifters eating at a deficit, where needs run higher because you're eating less overall. Reviews written for dieters who don't lift land lower, nearer 0.5 to 0.7 g per pound.
Our choice Ballpark suggests the lifter number to everyone.
The two errors aren't symmetric. Falling short while you're losing weight costs you muscle, and going over mostly costs you money, since a 2018 meta-analysis of 28 studies by Devries and colleagues found no effect of higher protein on kidney function in healthy adults. So the default leans high. If you don't lift, the lower number is defensible and the target is yours to edit. Targets round to the nearest 5 grams.
Our choice Above BMI 30, the protein calculation uses the weight at BMI 30 rather than your actual weight.
Protein needs track lean mass more closely than total weight, and Ballpark has no way to measure your lean mass. The alternative is asking you to estimate your own body fat percentage, which is a question almost nobody can answer accurately. Capping the basis weight is a rough approximation that avoids the question, and the approach follows recent work by Weijs and by Dekker and colleagues on calculating protein needs for higher body weights. It stays an approximation, and Ballpark says so where it appears.
No carb and fat targets
Published No protein-matched trial has shown an advantage for any particular carb and fat split.
The International Society of Sports Nutrition's 2017 position stand on diets and body composition puts it plainly:
With scant exception, all controlled interventions to date that matched protein and energy intake between ketogenic and non-ketogenic conditions have failed to show a fat loss advantage of the ketogenic diet.
The same paper quotes Kevin Hall: "There has never been an inpatient controlled feeding study testing the effects of isocaloric diets with equal protein that has reported significantly increased energy expenditure or greater loss of body fat with lower carbohydrate diets."
What separates these diets in practice is whether a person can stick to one, and you know that about yourself better than an algorithm does. Ballpark shows your carbohydrate and fat grams every day and assigns no target to either. If you're running keto or carb cycling, nothing here argues with you.
No micronutrient numbers
Published Most of those numbers would be invented.
U.S. labeling rules require only iron, calcium, vitamin D, and potassium beyond the main block of calories, macros, fiber, sodium, saturated fat, and sugars. Most manufacturers stop there. So a barcode scan reports nearly every other vitamin and mineral as zero. The food usually contains them and nobody was required to measure.
Real food doesn't hold still either. A 2018 study measuring retail produce across seasons found vitamin C in baby spinach ranging from 5.6 to 39.6 mg per serving, averaging 26 mg in winter against 11 mg in summer. Same vegetable and same store, with a sevenfold spread.
MacroFactor, who do track micronutrients and track them carefully, published their own assessment: content "can easily differ from label claims by 50% or more," and you can expect your real intake to differ from your logged intake "by up to ~25%." We agree with them, and we drew a different conclusion about whether to put those numbers on a screen.
Sources
Everything cited above, in full.
- National Academies of Sciences, Engineering, and Medicine. Dietary Reference Intakes for Energy (2023), Chapter 5. nationalacademies.org
- Shcherbina A, Mattsson CM, Waggott D, et al. Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. J Pers Med (2017). PubMed 28538708
- Mifflin MD, St Jeor ST, Hill LA, Scott BJ, Daugherty SA, Koh YO. A new predictive equation for resting energy expenditure in healthy individuals. Am J Clin Nutr (1990). PubMed 2305711
- Hall KD, Chow CC. Why is the 3500 kcal per pound weight loss rule wrong? Int J Obes (2013). PubMed 23774459
- Garthe I, Raastad T, Refsnes PE, Koivisto A, Sundgot-Borgen J. Effect of two different weight-loss rates on body composition and strength and power-related performance in elite athletes. Int J Sport Nutr Exerc Metab (2011). PubMed 21558571
- Ruiz-Castellano C, Espinar S, Contreras C, Mata F, Aragon AA, Martínez-Sanz JM. Achieving an optimal fat loss phase in resistance-trained athletes: a narrative review. Nutrients (2021). PMC8471721
- Iraki J, Fitschen P, Espinar S, Helms E. Nutrition recommendations for bodybuilders in the off-season: a narrative review. Sports (2019). PMC6680710
- Helms ER, Spence AJ, Sousa C, et al. Effect of small and large energy surpluses on strength, muscle, and skinfold thickness in resistance-trained individuals. Sports Med Open (2023). PMC10620361
- Garthe I, Raastad T, Refsnes PE, Sundgot-Borgen J. Effect of nutritional intervention on body composition and performance in elite athletes. Eur J Sport Sci (2013). PubMed 23679146
- Slater GJ, Dieter BP, Marsh DJ, Helms ER, Shaw G, Iraki J. Is an energy surplus required to maximize skeletal muscle hypertrophy associated with resistance training? Front Nutr (2019). PubMed 31482093
- Smith RW, Harty PS, Stratton MT, et al. Predicting adaptations to resistance training plus overfeeding using Bayesian regression: a preliminary investigation. J Funct Morphol Kinesiol (2021). PMC8167794
- Devries MC, Sithamparapillai A, Brimble KS, Banfield L, Morton RW, Phillips SM. Changes in kidney function do not differ between healthy adults consuming higher- compared with lower- or normal-protein diets. J Nutr (2018). PubMed 30383278
- Morton RW, Murphy KT, McKellar SR, et al. A systematic review, meta-analysis and meta-regression of the effect of protein supplementation on resistance training-induced gains in muscle mass and strength. Br J Sports Med (2018). PubMed 28698222
- Helms ER, Zinn C, Rowlands DS, Brown SR. A systematic review of dietary protein during caloric restriction in resistance trained lean athletes. Int J Sport Nutr Exerc Metab (2014). PubMed 24092765
- Weijs PJM. Protein requirement in obesity. Curr Opin Clin Nutr Metab Care (2025). PubMed 39514335
- Dekker IM, van Rijssen NM, Verreijen A, et al. Calculation of protein requirements: a comparison of calculations based on bodyweight and fat free mass. Clin Nutr ESPEN (2022). PubMed 35331517
- Aragon AA, Schoenfeld BJ, Wildman R, et al. International Society of Sports Nutrition position stand: diets and body composition. JISSN (2017). PMC5470183
- Phillips KM, Tarrago-Trani MT, McGinty RC, et al. Seasonal variability of the vitamin C content of fresh fruits and vegetables in a local retail market. J Sci Food Agric (2018). PubMed 29406576
- U.S. Food and Drug Administration. Nutrition labeling of food, 21 CFR 101.9. eCFR
- MacroFactor. Considerations for micronutrient tracking (2023). macrofactor.com
Last reviewed September 2026. If you find something on this page that misreads a source, or a study that should change one of these choices, we want to hear about it. Get in touch.
Track with a method you can read.
Ballpark is available on the App Store for iPhone.
