Beyond "60-90 g/h": why generic fueling rules fail

The problem with the generic range isn't that it's wrong — it's that it ignores the one variable that actually determines how much glycogen is being burned minute by minute: real relative intensity, expressed as watts against the cyclist's own FTP, not "how the ride feels."

Two riders covering the same route at the same average speed can have carbohydrate demands that differ by 40 g/h or more, simply because one weighs 15 kg more, has a different FTP, or their metabolic phenotype oxidizes fat differently. A fixed rule can't capture any of those three variables — a deterministic engine fed by real data can.

What follows is that calculation chain, link by link: from watts to kilojoules, from kilojoules to the grams of carbohydrate actually demanded, and from grams demanded to grams the gut can genuinely absorb.

It starts by converting the metric your power meter already measures — watts — into a real unit of mechanical work: the kilojoule. This is a physical conversion, not an estimate:

That mechanical work isn't directly metabolic expenditure — the human body doesn't convert chemical energy into movement at 100% efficiency. With a Gross Efficiency (GE) that in trained cyclists typically sits between 20% and 24%, total caloric expenditure is approximated as:

That's why RATIO never asks "how hard did that feel?" as its first input — it asks for watts. Real average power turns a subjective sensation into a verifiable kilojoule figure, the one input that feeds every calculation further down this chain.

Ref: Coyle, E.F. et al. (1992). Cycling efficiency is related to the percentage of Type I muscle fibers. Medicine & Science in Sports & Exercise, 24(7), 782-788; Moseley, L. & Jeukendrup, A.E. (2001). The reliability of cycling efficiency. Medicine & Science in Sports & Exercise, 33(4), 621-627.

Knowing total caloric expenditure isn't enough: you need to know what fraction is covered by fat (which the body carries in abundance) and what must be actively replaced with exogenous carbohydrate, which runs out within 60-90 minutes of moderate-to-high effort. That partitioning isn't linear — it follows the "Crossover Concept" described by Brooks & Mercier (1994): as relative intensity (%FTP) rises, there's a crossover point past which the fat oxidation pathway can no longer keep pace with ATP resynthesis demand, and the body increasingly shifts to glycolysis.

RATIO implements these bands deterministically: the carbohydrate oxidation rate isn't a fixed number per athlete — it's a stepped function of the %FTP actually sustained on each stretch of the ride. That's why a hard climb followed by a long descent doesn't demand the same carbohydrate as a flat ride of equal duration and average power — even though both would show the same figure in a Strava summary.

Ref: Brooks, G.A. & Mercier, J. (1994). Balance of carbohydrate and lipid utilization during exercise: the "crossover" concept. Journal of Applied Physiology, 76(6), 2253-2261.

Knowing how many grams of carbohydrate the body needs solves nothing if the gut can't absorb them at that rate. The intestinal epithelium transports glucose and maltodextrin exclusively through the sodium-dependent transporter SGLT-1, which saturates at around 60 g/h. Pushing past that figure with maltodextrin alone doesn't speed up absorption — it just accumulates undigested solute in the intestinal lumen, causing the gastrointestinal distress any cyclist recognizes past the two-hour mark.

The fix, documented by Jeukendrup (2004, 2014) and Jentjens & Jeukendrup (2005), is recruiting a second, sodium-independent transport pathway — GLUT-5, specific to fructose — that operates in parallel with SGLT-1 without competing for the same channel. Combining both sugars in the right ratio can raise combined exogenous absorption up to 90-120 g/h in athletes with trained gut adaptation ("gut training").

RATIO doesn't apply one fixed ratio: below 45 g/h, SGLT-1 alone isn't even saturated yet, so the mix is 100% maltodextrin — adding fructose there buys no extra absorption. Between 45 and 75 g/h, the engine switches to 2:1. Above 75 g/h, it shifts to 1:0.8 — the split dual-transporter research points to as optimal for absorption rates near the documented physiological maximum.

Ref: Jeukendrup, A.E. (2004). Carbohydrate intake during exercise and performance. Nutrition, 20(7-8), 669-677; Jeukendrup, A.E. (2014). A step towards personalized sports nutrition: carbohydrate intake during exercise. Sports Medicine, 44(S1), 25-33; Jentjens, R.L.P.G. & Jeukendrup, A.E. (2005). High rates of exogenous carbohydrate oxidation from a mixture of glucose and fructose ingested during prolonged cycling exercise. British Journal of Nutrition, 93(4), 485-492.

FTP is essential — it's the reference relative intensity is measured against — but treating it as the only physiological input is another mistake of the generic rule. Two riders with the same absolute FTP can have radically different oxidation profiles depending on their VLaMax (maximum lactate production rate): a "Diesel" phenotype has a more efficient fat oxidation pathway and burns less glycogen at low-to-moderate intensities, while an explosive "Puncher" phenotype shifts to glycolysis even at easy paces. RATIO models this with a multiplier — ×0.85 Diesel, ×1.00 Balanced, ×1.15 Explosive — applied only below 80% FTP, since above threshold every phenotype converges to the same near-exclusive glycolytic pathway.

The second limitation of FTP as a sole input is subtler: an average power figure can't tell a steady flat ride apart from a VO2 max interval session broken up by long recoveries. Both can share the same average power — but the second spends a meaningful fraction of its time well above that average, exactly where carbohydrate oxidation spikes toward the 100 g/h ceiling. Averaging hides precisely the data point that determines real glycogen expenditure.

That's why, whenever real power-zone data exists — from a structured workout or a post-ride analysis — RATIO doesn't use average power: it weights oxidation zone by zone against each %FTP band actually sustained, and only falls back to average power as an approximation when that per-zone data doesn't exist.

Ref: Allen, H. & Coggan, A. (2010). Training and Racing with a Power Meter (2nd ed.). VeloPress; Skiba, P.F. & Clarke, D.C. (2021). Cycling analytics and the modeling of physiological thresholds. Journal of Science and Cycling.

The "60-90 g/h" range isn't wrong as a population-level summary — but applied without watts, without kilojoules, and without phenotype, it's an eyeballed guess dressed up as a scientific figure. The difference between a rule of thumb and a deterministic engine isn't the literature each one cites — both can cite the same papers — it's whether that literature gets translated into code that responds, minute by minute, to the real power the cyclist is producing on that exact stretch of the ride.

That's the work RATIO's metabolic engine does every time a strategy gets calculated: it converts watts into kilojoules, kilojoules into substrate demand via the Crossover Concept, and that demand into a bottle ratio that respects real gut-absorption limits — without the cyclist ever having to memorize a single formula.

You can review every formula, threshold, and full bibliographic citation on the Methodology page.