The research behind the plan

The engine was built for mountain ultras first, and that is where most of these papers come from; the physiology of carbohydrate absorption, hydration and glycogen is the same on a road marathon or a long ride, and the targets scale with duration and effort. Every rule in the fueling engine points at one of the sources below. The plan screen in the app shows the same references next to each number under "Why these numbers". Where the evidence is thin, the app says so and calls the figure a working assumption.

Carbohydrate intake

• Tiller and colleagues, 2019. International Society of Sports Nutrition position stand on single-stage ultra-marathon. Sets the baseline ranges for carbohydrate, fluid, sodium and caffeine, documents how common gut trouble is, and describes the late-race shift to savoury food. The floor and ceiling of almost every target come from here.
• Jeukendrup, 2014. A step towards personalised sports nutrition: carbohydrate intake during exercise. About 60 g/h is the ceiling for a single sugar; glucose plus fructose lifts it towards 90 g/h. The gut is trainable, and intake should come down at low intensity. Why the target follows your tolerance band, not your weight.
• Costa and colleagues, 2019. Nutrition for ultramarathon running: trail, track and road. Around 90 g/h is supported for three hours or more but hard to tolerate in ultras; drink to thirst keeps hydration between the extremes. Why the plan tapers and why the fluid target is a range.
• Viribay and colleagues, 2020. 120 g/h in a mountain marathon cut muscle damage markers in gut-trained elites over 42 km. Treated as the upper bound, not a default: a lab result in trained guts over four hours, not an ultra recommendation.
• Costa and colleagues, 2017. Two weeks of gut training at 90 g/h during two-hour runs cut symptoms by about 60 percent. Why the plan refuses to prescribe a rate you have not practised, and why a clean long run can raise your tolerance band.
• Stellingwerff and Cox, 2014. Systematic review of carbohydrate supplementation and performance. Benefit grows with duration; over two hours the mechanism is delivery rate. Why multiple transportable carbohydrates matter above 60 g/h.

Gut trouble

• Costa and colleagues, 2025. Sports Dietitians Australia and Ultra Sports Science Foundation practitioner guide to exercise-associated gastrointestinal symptoms. Moderate to high carbohydrate intake protects the gut lining; fructose is the main culprit at high doses; dehydration and forced drinking both make symptoms worse. Why a history of nausea caps the target and shortens the interval.
• Stuempfle and Hoffman, 2015. Gastrointestinal distress at a 161 km race: 96 percent of finishers had symptoms, worst in the hottest section, and gut trouble was the reason to drop for a third of those who did. Why heat reduces the target.
• Stuempfle and colleagues, 2013. Runners without gut distress drank more and ate more fat; low fluid intake came before the symptoms. Why fluid has a floor.

Fluids and sodium

• Baker and colleagues, 2016. Normative data for sweat sodium and sweating rate in athletes: sodium around 830 mg per litre with a wide range, sweat rates from a quarter of a litre to several litres an hour. Why sodium is planned per litre of sweat rather than as one number, and why a "salty sweater" flag exists.

The tank: the glycogen estimate on the watch

• Romijn and colleagues, 1993. Fat and carbohydrate oxidation across exercise intensities. Carbohydrate's share of energy rises with intensity. The watch burns each second's calories at a share read from your heart rate along this curve.
• Achten, Gleeson and Jeukendrup, 2002. Maximal fat oxidation sits around 56 to 64 percent of aerobic capacity in trained athletes. Ultra pace lives there, so about half your fuel is carbohydrate at an easy shuffle.
• Coyle and colleagues, 1986. Glycogen use fed and unfed at 71 percent of aerobic capacity. Ingested carbohydrate spares muscle glycogen almost one for one, and fatigue unfed followed low blood glucose. Why confirmed gels go straight back onto the store.
• Coyle and Gonzalez-Alonso, 2001. Cardiovascular drift: heart rate creeps up at constant work after twenty minutes, largely from dehydration. Why the watch allows for drift late in a long race instead of reading it as harder effort.
• Endocrine Reviews, 2026. Review of 160 studies on carbohydrate ingestion, metabolism and performance. Fatigue tracks falling blood glucose more closely than glycogen depletion itself. Why the hourly intake, not the tank, is the biggest number on the screen.

Starting glycogen

The tank starts at a value scaled to body mass: 7.5 g per kg after a race-week carb load, 6 on an ordinary day, about 5.4 fasted, since an overnight fast drains roughly half of the liver's 80 to 100 g while muscle stores are untouched (Nilsson and Hultman, 1973; Bergström and Hultman on muscle glycogen). The app calls this a working assumption and lets you set an exact number.

The runner's own data

Alongside the literature, the rules were checked against one ultra runner's logged races: a 48 km mountain race fuelled at about 73 g/h for ten hours with no gut trouble, a race lost to a protein drink mix, and a hand-built 80 km plan the engine had to reproduce before it was trusted. Those are private records and are not published; they are the reason the engine treats protein during a race as a hard exclusion and front-loads the gels.

General guidance from published research and the data you enter. Not medical advice. Rehearse any plan in training and listen to your body over any prompt.