Calorie Tracking for Athletes: Fueling a Real Training Load

Almost everything written about calorie tracking assumes you are trying to eat less. Calorie tracking for athletes is a different job. The people who need it most are usually eating too little for the work they are asking their body to do, and the app tells them they are doing great, because the app was built to reward a low number.

The fix isn't more precision. It's changing which number you look at.

Energy availability, not your deficit

The metric sports nutrition uses is energy availability: what's left of your intake once training has taken its share, divided by your fat-free mass.

Energy availability = (calories eaten - exercise calories burned) / kg fat-free mass

Take a 70 kg athlete at roughly 15% body fat, so about 59.5 kg of fat-free mass. She eats 2,800 calories and burns 700 in training. That leaves 2,100 calories to run everything else, which works out to about 35 per kg of fat-free mass.

Nothing about that day looks alarming on a food log. A 2,800-calorie intake reads as generous. The point of the calculation is that the training bill gets paid first, and what's left over is what your endocrine system, immune system, and bone remodeling have to share.

Loucks and Thuma ran the experiment that put a number on this. They held 29 regularly menstruating women at set energy availabilities for five days and watched luteinizing hormone pulsatility. At 30 kcal per kg of lean body mass per day, pulsatility held. Below that, pulse frequency fell.

That study gets quoted as though 30 were a universal red line, and it isn't. The subjects were habitually sedentary women, the exposure was five days, and the outcome was one hormone.

The 2023 IOC consensus statement on Relative Energy Deficiency in Sport backed away from a single cutoff on purpose. It distinguishes low energy availability a body adapts to from the kind that causes problems, moderated by individual factors, and counts more than 170 original research papers since the 2018 version, including new data on male athletes. Its argument against one clinical threshold is that the level which matters varies by individual, by sex, and by which body system you're asking about.

So treat the number as a smoke alarm, not a diagnosis. If your own arithmetic keeps landing in the low 30s or under while training volume climbs, that's worth investigating rather than optimizing.

Your training burn is the shakiest term in the equation

Energy availability has two inputs and both are estimates. The exercise one is worse than most people assume.

Stanford researchers put seven wrist-worn devices against indirect calorimetry across sitting, walking, running, and cycling in 60 volunteers of varied age, size, and fitness. Six of the seven held median heart rate error under 5% during cycling. Not one got energy expenditure error below 20%. The best performer in the group was still off by more than a fifth.

MET-based estimates, which is what a calculator uses when you tell it what you did and for how long, have a different problem. They're gross figures. They include the calories you would have burned sitting on the couch during that same hour, so subtracting them from your intake double-counts your resting metabolism.

Here's the size of that at 155 lb (70 kg), using the MET values behind the exercise pages:

Activity	MET	Gross per hour	Net after resting burn
Running, 6 mph	9.8	689	619
Cycling, moderate	8.0	562	492
Swimming, moderate	7.0	492	422
Weight lifting	3.5	246	176

For running, the correction is small enough to ignore. For lifting it removes 28% of the number, which is why strength athletes who add their training calories back at face value often can't work out where their surplus went. Most of a lifting session is spent standing around between sets, which is barely above sitting still.

If you're deciding whether to eat those calories back at all, we went through the case both ways in should you eat back exercise calories. The short version for athletes: yes, but from a planned training-day target rather than from whatever a watch reported that afternoon.

Protein: what the position stands actually say

Two documents carry most of the weight here, and they mostly agree.

The ACSM, Academy of Nutrition and Dietetics, and Dietitians of Canada joint position statement puts athlete protein needs at 1.2 to 2.0 g per kg per day. The ISSN position stand on protein and exercise lands at 1.4 to 2.0 g/kg/day for building and maintaining muscle mass, and adds two useful details: 2.3 to 3.1 g/kg/day may be needed to hold onto lean mass during a calorie deficit in resistance-trained people, and per-meal doses of roughly 0.25 g/kg, or 20 to 40 g, spread every three to four hours.

For a 70 kg athlete at 1.6 g/kg, that's 112 g a day. Whole foods get you there without much drama:

Food, 100 g cooked	Protein	Fat
Chicken breast	31 g	3.6 g
Ground beef	24 g	11.2 g
Salmon	25.4 g	8.1 g

The gap between the first two rows is the whole argument in chicken breast vs ground beef. Same category, seven grams of protein apart, three times the fat. Neither is wrong. On a heavy training day the beef's extra calories may be exactly what you need, and on a weight-restricted day they're what you don't.

Weigh your protein rather than eyeballing it, and stay consistent about whether you weigh it raw or cooked. Meat is the one place where a 30 g estimation error moves a macro target and not just a calorie count.

Carbohydrate scales with the session, not with the day

This is where athlete tracking diverges hardest from general advice. The joint position statement sets daily carbohydrate targets against training load:

Training load	Carbohydrate
Light	3 to 5 g/kg/day
Moderate, about 1 h/day	5 to 7 g/kg/day
Endurance, 1 to 3 h/day at moderate to high intensity	6 to 10 g/kg/day
Extreme, more than 4 to 5 h/day at moderate to high intensity	8 to 12 g/kg/day

In that top band, a 70 kg athlete starts at 560 g of carbohydrate and can be asked for 840. Put that in food terms and the problem becomes obvious: cooked white rice carries 28.2 g of carbohydrate per 100 g, so 560 g is roughly two kilograms of rice a day, before anything else on the plate.

That's why the food choices that serve a cutting phase can work against you here. Sweet potato vs rice normally goes to the sweet potato: 90 calories per 100 g against 130, with 3.3 g of fiber against 0.4. Flip the goal to hitting 560 g of carbohydrate without a heavy gut on race morning, and rice's lower fiber and higher carb density stop being drawbacks.

During long sessions the guidance is per hour rather than per day. Burke and colleagues put 30 to 60 g per hour as the appropriate target for sports of longer duration, with events past about 2.5 hours potentially benefiting from up to 90 g per hour using blends of different carbohydrate types. For sustained high-intensity efforts around an hour, they note that very small amounts, including a mouth rinse, can help through the central nervous system rather than through fuel.

Log those in-session calories. A gel every 30 minutes over a four-hour ride is real food that most people never enter.

How to track a training week

Daily precision is the wrong target. Training loads swing across a week, appetite often lags behind effort, and a Tuesday deficit followed by a Thursday surplus is a normal week rather than a problem.

1. Set a baseline maintenance figure and two training-day targets, one for easy days and one for hard days. Two numbers you'll actually hit beat one you keep missing.
2. Weigh protein and fats. Skip weighing vegetables. Estimation error scales with calorie density, so the accuracy you buy from weighing spinach is close to nothing.
3. Log training as a planned session, not as whatever your watch decided afterwards. Use the same estimate every time you do the same workout so a change in the number means a change in the work.
4. Check the week's total, not the day's. Seven days smooths out the noise from digestion, sodium, and glycogen without hiding a real trend.
5. Recheck body mass and performance every two weeks. If training quality drops while the log looks fine, the log is wrong.

Step 5 is the one people skip. Your log is a model of what you ate, and it gets less accurate the longer you go without checking it against something outside the app.

Athletes under-report worse than they think

A systematic review of dietary assessment in athletes found 11 studies comparing self-reported intake against energy expenditure measured by doubly labelled water, the most accurate method available. Pooled, athletes under-reported intake by 19%, an average of about 2,793 kJ a day, which is roughly 670 calories. The pooled effect size was large.

That's a meal a day, unaccounted for. The shortfall is not usually deception. It's the oil in the pan, the second helping, the mid-ride gel, the handful of something on the way out the door.

For someone dieting, a 19% under-report means slower progress than the math predicted. For an athlete it means the energy availability calculation you just did is optimistic, and the direction of the error is the dangerous one. If your numbers say you're fueling adequately and your body says otherwise, believe your body.