Fuel

Fuel: A Training Nutrition Plan That Moves With Your Actual Training Load

Cameron Usher · Founder, px

1 October 2026 · 19 min read

Twelve years in competitive sport, trained alongside the Australian Olympic squad. px is the coach he needed after ten coaches, every wearable and every protocol.

You logged everything yesterday. Oats by barcode, lunch by photo, the protein bar in the car park. You also ran intervals at six, slept five and a half hours and sat through a meeting that ran two hours over. At the end of the day you were told you had hit your number. It was the same number you get on a rest day.

A training nutrition plan works only when it moves with the day you actually had. Carbohydrate needs rise and fall with training load. Protein works best as a steady daily total, spread across meals. And a food log changes nothing until something reads it against today's training, sleep and recovery. Below is the evidence for each, the myths it clears out, and what it looks like when food becomes the third input to one daily plan, not a separate lane.

Why does logging your food change so little on its own?

Because a food log compares a soft estimate of what you ate with a number that does not move. Each half of that sentence has been measured.

The numbers going in are softer than they look

Checked against doubly labelled water, the reference method for what the body actually spends, food records come up short. Across 11 such studies in athletes, Capling and colleagues (2017, Nutrients) found intake under-reported by 19 per cent on average, ranging from almost nothing to 36 per cent. The studies were small, 7 to 19 athletes each, but the direction was consistent.

Technology does not escape it. Across 59 studies and 6,298 adults, Burrows, Ho and Rollo (2019, Frontiers in Endocrinology) found most showed significant under-reporting, and technology-based methods came in 6 to 37 per cent under. A photo is faster to log. It is not automatically truer.

The reference numbers are soft too:

• Labels carry a legal margin. European Commission guidance on label tolerances (2012) allows protein, carbohydrate and sugars declared at 10 to 40 g per 100 g to sit 20 per cent either side. A label reading 20 g of protein per 100 g is compliant anywhere from 16 to 24 g.
• Energy values are a model. In a controlled feeding study of 18 adults, Novotny, Gebauer and Baer (2012, American Journal of Clinical Nutrition) measured 129 kcal from a 28 g serving of almonds. The standard calculation says 168 to 170. That is a 32 per cent overestimate for one ordinary food.
• Databases drift. When a dietetics-trained researcher entered 30 one-day food recalls into five popular tracking apps, Griffiths, Harnack and Pereira (2018, Public Health Nutrition) found the apps tended to underestimate, by 7 to 41 per cent across nutrients, and two of the five significantly underestimated protein.

None of this makes logging pointless. It makes the number at the bottom of the day an estimate with a wide margin. A plan that takes it as exact will be precisely wrong.

Logging fades faster than the need for it

In a randomised trial by Laing and colleagues (2014, Annals of Internal Medicine), 105 adults in primary care were given a free calorie-counting app for six months. Ninety-four logged in during the first month. By month six, 34 did, and the median had fallen from eight logins a month to zero. The app on its own made no significant difference to the outcome the trial measured. They were patients, not athletes, but the Mind pillar on staying consistent with training found the same fall-off in training apps.

Your appetite does not read your training either

Hunger does not close the gap on a hard day, at least not the same day. A meta-analysis of 29 studies and 51 trials by Schubert and colleagues (2013, Appetite) found that people barely changed what they ate in the hours after exercise (effect size 0.14, trivial), so their intake relative to what the session cost fell sharply (effect size minus 1.35). In the authors' words, people "tend not to compensate". These were lab meals, not whole days. Hunger is a slow instrument, and a hard morning can leave you a long way short before you feel it.

So the target is the problem

Put those together. An imprecise log is checked against a fixed number by someone whose appetite is not tracking the day. The fixed number is what breaks. Your needs on a rest day and on a two-hour day are not close, and nothing in the log knows which day it was.

The Mind pillar showed that self-monitoring works best alongside other techniques. For food, the missing one is obvious. Something has to read the log against the day.

How much should your eating change with your training load?

More than a fixed target allows, and mostly through carbohydrate. The clearest statement is the joint position of the Academy of Nutrition and Dietetics, Dietitians of Canada and the American College of Sports Medicine (Thomas, Erdman and Burke, 2016, Medicine and Science in Sports and Exercise). Here are its ranges for one illustrative 70 kg person across four kinds of day.

Ranges from Thomas, Erdman and Burke (2016). The 70 kg column is arithmetic for illustration, not a prescription.

Two things stand out. Carbohydrate is the line that moves, about two and a half times from the lightest row to the heaviest, midpoint to midpoint. Protein holds.

Be honest about where you sit. If you train 45 minutes on most days, you live in the top two rows. The bottom rows belong to people training for hours. What matters is not the size of your swing. It is that there is one, and a single daily number cannot follow it.

Fuel for the work required

Impey and colleagues (2018, Sports Medicine) argue that carbohydrate availability "should be manipulated day-to-day and meal-by-meal according to the intensity, duration and specific training goals" of the session in front of you. More before and around the hard work. Less around the easy work. The authors are clear that this is a theoretical framework with limited long-term evidence, which is worth knowing before anyone sells it to you as a protocol. The principle underneath it, that fuel should follow the work, is not in dispute.

Are carbs bad on rest days?

No. A lighter day needs less carbohydrate, not none. The joint position stand has no separate rest-day row, and its lightest category still sits at 3 to 5 g per kg. The day after hard training is also where the next session gets fuelled. Burke, van Loon and Hawley (2017, Journal of Applied Physiology) describe the later phase of recovery, from about 4 to 24 hours, as the time when intake should meet the fuel needs of the training to come. Friday's pasta is Saturday's long run.

What about training without breakfast?

Training fasted is a legitimate choice for some sessions and a poor one for others. Our guide to fasted versus fed training sets out the evidence: easy or short work done fasted costs you little, long or hard work done fasted costs you quality, and deliberate low-carbohydrate training as a tool for adaptation is still an open question. Burke and colleagues (2011, Journal of Sports Sciences) put it plainly: whether these "train-low" strategies improve performance in well-trained people "is unclear".

How much protein do you need, and does timing matter?

A daily total inside a fairly narrow band, spread across the day. Timing matters far less than the internet suggests. Position stands give ranges, and so will we. Your own number depends on your body, training and goal, which is why it should come from your data, not an article.

The daily range

• The joint position stand puts protein at 1.2 to 2.0 g per kg per day for athletes, with the higher end for heavy training blocks (Thomas, Erdman and Burke, 2016).
• The International Society of Sports Nutrition says 1.4 to 2.0 g per kg per day is sufficient for most exercising people (Jäger and colleagues, 2017, Journal of the International Society of Sports Nutrition).
• The strongest single analysis is a meta-analysis of 49 randomised trials and 1,863 people doing resistance training (Morton and colleagues, 2018, British Journal of Sports Medicine). Gains in lean mass levelled off at 1.62 g per kg per day, with a wide confidence interval of 1.03 to 2.20. Because of that upper bound, the authors suggest about 2.2 g per kg may be the prudent figure for someone trying to get the most from strength training.

Spread beats piled up

How you split the total matters a little. In a small, careful study of 24 trained men (Areta and colleagues, 2013, Journal of Physiology), 80 g of whey over 12 hours of recovery was given as eight 10 g doses, four 20 g doses or two 40 g doses. Four 20 g doses produced 31 to 48 per cent more muscle protein synthesis than either alternative. It is one acute study in men using protein drinks, but it fits the ISSN's advice of protein every 3 to 4 hours and the case made by Schoenfeld and Aragon (2018, Journal of the International Society of Sports Nutrition) for at least four protein-containing meals. In plain terms: protein at breakfast, not just at dinner.

The two timing myths

"You can only absorb 30 g at a time." Absorption is, in Schoenfeld and Aragon's words, "virtually unlimited". The old 20 to 25 g figure describes how much fast-digesting protein on its own maximally switches on muscle building in one sitting, not what the gut can take in. A 2023 trial of 36 recreationally active young men (Trommelen and colleagues, Cell Reports Medicine) found 100 g produced a larger and longer anabolic response, over more than 12 hours, than 25 g. A commentary by Witard and Mettler (2024) notes the trial had no spread-out arm and was not in trained lifters. Read it as "the ceiling is not where you were told", not as a reason to eat it all at once.

"You have 30 minutes after training." A meta-analysis of 23 studies (Schoenfeld, Aragon and Krieger, 2013, Journal of the International Society of Sports Nutrition) found the apparent benefit of protein within an hour of training disappeared once total daily protein was accounted for. The ISSN stand describes the anabolic effect of exercise as lasting at least 24 hours. Most of those studies were in untrained people, so the finding is firmer for beginners than for seasoned lifters.

The timing rule that is real

There is a clock that matters, and it belongs to carbohydrate. When two demanding sessions sit less than 8 hours apart, the joint position stand advises rapid refuelling, and Burke, van Loon and Hawley describe the first four hours as the phase where prompt carbohydrate does the most. If the next hard session is tomorrow, total intake across the day matters more than the minute you eat.

The morning is where most people meet this question. Our guide to whether you need protein before a morning workout answers it properly: usually not as a rule, because the daily total and its spread do more than a pre-dawn shake, and before a long or hard morning session carbohydrate matters more than protein.

What does under-fuelling cost your training?

The adaptation you trained for. When intake leaves too little energy for the body's needs after training has taken its share, sport scientists call it low energy availability. The International Olympic Committee's 2023 consensus statement (Mountjoy and colleagues, British Journal of Sports Medicine) separates a short, mild shortfall that the body adapts to and reverses quickly from a prolonged or severe one that does not. Among the performance costs of the second kind it lists a weaker response to training, lower endurance, and lower strength and power.

It was written for competitive athletes, so take the principle rather than the magnitude. Training is the stimulus. Food is part of what turns it into fitness. Under-eat the days you train hardest and you pay for the session without collecting all of the return.

It is rarely deliberate. Remember Schubert's finding: appetite does not step up on the same day. A morning session, a working lunch that did not happen and a late dinner can leave a training day badly short without a single decision to eat less. The Move pillar on biometric training makes the same point from the training side: under-eat a heavy training day and your recovery pays for it that night and the next day.

This is why px approaches food as permission and plan, not restriction. Eating enough on a hard day is part of the training, not a reward for it, in the same way that rest is not a reward for work done.

The same logic applies when training drops for a reason you did not choose. Our guide to nutrition for injury recovery covers the comeback in full. The short version, from a review by Tipton (2015, Sports Medicine): energy balance is critical, deficiencies should be avoided, and even if total food comes down with the lighter week, protein should not. Anything clinical belongs to your physiotherapist or doctor.

Why do sleep, stress and heat change what you need?

Because the body keeps one account. Sleep, work and weather all draw on the same recovery that training does, and all three push eating around.

Sleep

A meta-analysis of 11 studies by Al Khatib and colleagues (2017, European Journal of Clinical Nutrition) found that after partial sleep loss people ate about 385 kcal more a day (95 per cent confidence interval 252 to 517), with no change in the energy they spent. The mix shifted too: more fat and less protein. There was no pooled effect on carbohydrate, which is worth knowing, because "tired means sugar" is the popular version. These were short laboratory studies.

The Move side already says it: do not under-eat the tired day. The useful move is to plan it. Keep protein up, because the pooled data show it slipping, keep meals regular, and do not read the extra hunger as a failing. Craving and eating are also different measurements. Our guide to why sleep deprivation drives food cravings goes through what a short night does to hunger and food choice, and how to plan the morning after.

Stress

A hard week at work suppresses the same recovery signals as a hard week of training, as the Mind side shows in how work stress shows up in your HRV. It moves food too. A meta-analysis of 54 studies and 119,820 healthy adults (Hill and colleagues, 2021, Health Psychology Review) found stress nudged overall intake up and shifted what people chose. The effects were small and varied a lot from study to study: a tendency, not a rule. px's coaching rule for a heavy week is regular meals and no restriction. A stressful fortnight is the worst time to start a tighter plan.

Heat

Heat raises the cost of the same session. The Move side notes that an hour of warm-weather training can cost a litre or more of sweat, and hands the fuelling question to this side of the site. Valtin (2002, American Journal of Physiology), the review that found no scientific basis for "eight glasses a day", was careful to add that larger intakes are advisable during vigorous work and exercise, especially in hot climates. Our guide to hydration and nutrition for heat training covers fluid, salt and eating on the road: start level, bring sodium into long hot sessions rather than water alone, and keep fuelling steady when travel breaks your routine.

Which nutrition rules can you safely drop?

Most of the ones that arrive as absolutes. Here are five, and what the evidence says instead.

Yamada and colleagues (2022, Science) measured water turnover in 5,604 people across 23 countries and found it varied with body size and composition, physical activity, athletic status and climate, among other things. One number for everyone cannot be right. Thirst, the session and the weather are better guides.

Why flexible holds and rigid breaks

Rules like these persist because they feel like control. The evidence on control points the other way. In a study of 223 adults by Smith and colleagues (1999, Appetite), flexible control of eating went with less overeating, while rigid control, strict calorie counting included, went with more. It is correlational and it was not in athletes, so it cannot prove cause. Rigid rules also leave no room for the day that does not fit, and a training week is full of those. Our guide to whether clean eating is necessary takes that argument apart properly. The short answer is no. What matters is intake matched to your training and a pattern you can hold on a bad week.

The rules worth keeping are practical. On a packed week the problem is rarely recipes. It is decisions. Our guide to quick meals for a busy training schedule builds a short rotation of fast meals, each with a protein anchor and a carbohydrate portion that grows or shrinks with the day. It pairs with the Move side's training plan for busy professionals: two to three sessions a week, fuelled by food that fits the same tight week.

What does a day look like when food reads your training?

Here is an ordinary training day, with food as one of three inputs to the same plan.

Before you are up, the day's briefing is ready. It has four cards: your overall px Score, Move, Mind and Fuel, each with one action and the reason for it.

You run at six. The session comes in from your watch, and the target px reasons against when you ask it what to eat rises with the training you have actually logged, up to a cap, unless your goal is a cut. The carbohydrate and fat in that target scale with it. If your wearable reads it as a high-load day, the protein in that target goes up too. The daily target in Your Nutrition follows your preferred training days: a little more on a training day, a little less on a day off. Less, not none.

You log breakfast with a photo and confirm the items px reads from it. That is one of five ways in, with search, barcode, quick log and simply telling Ask px. None is exact, for the reasons above, which is why px reads the log as an input to the day rather than a verdict on it.

At lunch you open the recommended day. It is built from what is still left to eat today, not from a fixed menu. Every main meal carries a protein anchor, and a row might tell you it covers your last 28 g of protein. Your allergens and your diet are hard filters, so nothing that breaks them is offered.

If you slept badly, the day accounts for it. If your goal has you eating under maintenance, a thin night halves that gap for the day and Your Nutrition says why. If your recovery has collapsed, the recipes lean towards ones that support recovery and the cooking time is capped. Nobody wrecked by the week needs a fifty-minute recipe.

At seven you cook in cook mode. Step timers keep counting with the phone locked, you can ask to swap an ingredient you do not have, and the recipe scales to however many you are feeding.

Water sits beside it all as a guide scaled to your bodyweight, never pushed past thirst. On a long or hot session, px tells you to add electrolytes rather than water alone.

Now the part where the three pillars meet. Take a different day. The run was hard, lunch was a coffee and the evening is getting on. You are well under target on a heavy training day. px does not scold you. It changes the plan:

"You are well short of what this morning's run cost and the evening is getting on, so tonight's session is ten minutes shorter rather than drawing on reserves you have not replaced."

That is what Direction looks like for food. The log did something. It changed what you do next, and it told you why. Why that only works when one system reads all three is the case made in why an all-in-one fitness coach beats ten apps.

What px does not do matters too. Without your body data, it shows you how to set a target rather than inventing one. It does not place a snack before your session automatically, so ask px what to eat before a 7am start and it answers on the principles set out here. And when you want a human, it hands you to a person on the px team.

Your food log should be able to change your training. Here, it does. px reads what you already track and builds one plan across training, mind and food before your day starts. Try px

You have read how it thinks. Now let it coach you.

Seven days free. Cancel any time.

Start free

← Back to px Insights