Nutrition

Fat

20–35% of daily energy

Below that, there's a real but contested link to lower testosterone — the full picture is under Fat, below — and that's the reason for this floor, not a round number. Above it, once protein and carbohydrate needs are met, there's no evidence more fat does anything further.

Evidence: solid for the range itself — a standing dietary-reference figure (Institute of Medicine, 2005), not a single study. The reasoning behind the floor is more contested; see Fat, below.

Carbohydrate

3–10 g/kg/day, scaling with training load

3–5 g/kg/day covers light or skill-based training. 5–7 g/kg/day covers a normal training day, like most sessions in this programme. 6–10 g/kg/day covers genuinely high-volume days. Carbohydrate is what glycogen is made from, and glycogen — not just hunger — is what runs out during a longer or harder session.

Evidence: strong — a standing joint position stand from the three major sports-nutrition bodies (Academy of Nutrition and Dietetics, Dietitians of Canada, American College of Sports Medicine), not a single study.

Cholesterol and testosterone

Adequate matters. More doesn't.

Cholesterol is the actual biological starting material testosterone is built from — that part is settled biochemistry, not a supplement-industry talking point.

It doesn't follow that eating more fat or more cholesterol raises testosterone. It doesn't, once you're already eating enough. What the evidence more consistently points to is the other direction: very low fat intake — well under the 20–35% range above — is linked to lower testosterone in some studies. That link isn't fully settled either: the newest and largest analysis on this exact question found no effect at all, and large population studies haven't found an independent link once other factors are accounted for.

Usable version: don't go very low on fat. Once you're eating enough — the range above — there's no evidence more fat or more dietary cholesterol does anything further for testosterone specifically. Good fat sources matter for other reasons (below); testosterone past adequacy isn't one of them.

Evidence: the precursor biochemistry is settled. The "low fat lowers testosterone" claim is genuinely contested — a smaller, older meta-analysis found a real drop; the newest, nearly four-times-larger meta-analysis on the identical question found no effect at all; large observational cohorts found no independent link once adjusted. Treat "eat enough fat" as sound and the specific numeric threshold as less certain than it's often presented.

Fat quality and food processing

Whole foods, less processing

Good fat sources: olive oil, avocado, fish, eggs. None of these are controversial on their own merits.

Minimise ultra-processed food. This is one of the best-supported claims on this page: a review of 45 pooled analyses across almost 10 million people found convincing evidence linking it to higher cardiovascular mortality, type 2 diabetes, and anxiety. It's observational evidence, not a trial — nobody can ethically run a long-term trial that deliberately feeds people a harmful diet — but the association is large and consistent.

What this page won't say: that seed oils specifically are harmful. That claim is popular, but the weight of trial evidence actually points the other way — replacing saturated fat with the polyunsaturated fat in most seed oils modestly lowers cardiovascular risk in the best-controlled studies available (a Cochrane review, and the American Heart Association's own position). The one well-known study behind the "seed oils are harmful" claim is a single reanalysis of a 50-year-old trial using a specific 1960s corn-oil formulation — a real finding, but not enough to outweigh the newer, larger evidence.

The honest version of "avoid the bad stuff": eat whole-food fat sources and less industrially processed food in general. That's well supported. Blaming one specific ingredient in it isn't.

Evidence: ultra-processed food — strong, consistent, large-scale association, not proof of causation, and it can't ethically become one. Seed oils specifically causing harm — contested, and the current weight of randomised evidence runs against it. Good-fat sources — uncontroversial on their own merits.

Brain fat

~50–60% fat, by dry weight

The brain is roughly 50–60% fat by dry weight, once the water is accounted for — one of the most fat-dense tissues in the body, and heavily enriched in DHA, an omega-3 fatty acid that's a real, structural part of brain-cell membranes.

What this doesn't mean: that eating more fat sharpens your thinking, or balances your hormones through the brain specifically. Hormone production runs on cholesterol elsewhere in the body (see Cholesterol and testosterone, above) — a separate mechanism from what the brain's own fat content does. And the trial evidence for omega-3/DHA improving cognitive performance is genuinely mixed: benefit shows up most consistently in people who start with a measurable decline or deficiency, not reliably in a healthy, adequately-fed adult. This page won't promise a strong mind from fat intake — DHA earns its place here as a structural fact, not a performance claim.

Evidence: the dry-weight figure and DHA's structural role are solid, standard neuroscience. The cognitive-benefit literature is mixed and dose/population-dependent — most consistent evidence is in people correcting a deficiency or age-related decline, not in healthy adults generally.

An accidental short gap

Eat 2–3 hours before training. No exceptions.

Training on an empty stomach before a normal session carries a real, measurable cost: pre-exercise carbohydrate availability reliably improves how many reps and sets get completed once a session runs past about 45 minutes, and the cost shows up fastest at genuinely low energy — a flat, under-fuelled day is exactly when a working set gets cut short. Eating a real meal with protein and carbohydrate 1–3 hours beforehand removes that cost. Of everything on this page, this is the simplest, highest-value rule.

Evidence: solid — pre-exercise carbohydrate reliably improves reps/sets completed over sessions longer than ~45 minutes; long-run strength and muscle differences between fasted and fed training are small when it's consistent over weeks. The cost is to that day's session, not necessarily to months of progress.

A deliberate multi-day fast

If you fast: training through it is possible. Expect a real cost, not a free pass.

Training deep in a multi-day water fast is a real, sometimes-successful practice — the evidence doesn't say never do it. But by 72 hours, glycogen is genuinely depleted, and the muscle-building response to a hard set is largely gone: the post-exercise anabolic signalling that normally rises after a set doesn't rise at all after severe fasting. Plan for a lower target RPE, a shorter session, and technical or moderate-load work — not a new max — and treat a strong session during a fast as a good outcome, not the one to expect every time.

Evidence: mixed, but each part solid on its own — the acute performance cost and the blunted post-exercise signalling are both real, measured effects. A single good fasted session, even one near a lifetime best, is a genuine outlier, not proof the cost isn't there.

Hydration and electrolytes

Fasting lowers insulin, which increases how much sodium you lose through urine — training sweat adds to that. On any fast past 24 hours, add salt to your water (or a sugar-free electrolyte source), more on a day you train, and drink to thirst plus extra to cover sweat. Never take a large one-off dose of potassium — sip small amounts through the day instead.

Evidence: practice-based — drawn from supervised prolonged-fasting protocols, not randomised trials. Treat these as reasonable defaults, not a clinically-graded prescription.

Creatine during a fast

Creatine carries no meaningful calories and doesn't trigger a real insulin response at a normal 3–5 g dose, so it doesn't work against the metabolic aims most people fast for. Keep taking the daily 5 g through the fast — daily consistency matters far more than when any single dose gets absorbed. If you want a small, genuine edge in uptake, pairing that dose with the first carbohydrate-containing meal at the end of the fast is real, but minor — not a requirement. A strict "water only" definition of a fast would still count any dissolved creatine as breaking it — that's a personal, definitional choice, not a safety or efficacy question.

Evidence: a real, replicated mechanism (insulin increases creatine uptake into muscle) — the practical consequence for a fasting lifter is a minor optimisation, not a change to the 5 g/day advice above.

Refeeding

Break a fast gradually over the next day — don't end a multi-day fast with one large meal. Restart the every-3–4-hour protein pattern from the first refeed meal, but ease into it: moderate portions for the first 24–48 hours rather than a full 30–40 g leucine-rich meal cold. Keep electrolytes a priority through the first one to two days of refeeding, not just during the fast itself — that's the window rapid electrolyte shifts actually happen in.

A 72–96 hour fast sits below where refeeding risk is usually seen clinically — typically well beyond a week, and in malnourished populations. That's not a reason to treat a fast of this length as risky by default, but it is a reason to always refeed gradually, and to take unusual weakness, swelling, breathlessness, or heart-rhythm changes seriously in the days after breaking a fast.

Evidence: a careful extrapolation — the clinical refeeding-syndrome literature concerns much longer or malnourished fasting than this. The principle (refeed gradually, watch electrolytes) is sound; the specific risk thresholds come from a different population.