The Protocol

Dossier · goal-specific

Programs

Hypertrophy is one adaptation among several, and the others are not obtained by doing hypertrophy training harder. Sprinting is a horizontal-force problem. Marathon running is a fractional-utilisation and economy problem. Jumping is a rate-of-force-development and tendon-stiffness problem. Mobility is a neural-tolerance problem before it is a tissue-length one. Each below is built from the people who actually study it, with the mechanism first and the programme second — and with the interference between them stated honestly rather than wished away.

Research vintage · Q3 2026

Read first

Why you cannot run all four at once

RCT / meta-analysisPrimary literature

The interference effect, mechanistically

Mechanism

AMPK versus mTORC1 — a genuine molecular conflict

Prolonged endurance work depletes ATP relative to AMP, activating AMPK via LKB1. AMPK then does two things that directly oppose hypertrophy: it phosphorylates TSC2 (activating it, which switches Rheb off and therefore mTORC1 off) and it phosphorylates Raptor directly, inhibiting mTORC1 assembly. Simultaneously AMPK drives PGC-1α, the master regulator of mitochondrial biogenesis. So the same molecular switch that builds an aerobic engine actively suppresses the anabolic signal you lift for. This is not a metaphor — it is one kinase with two opposing downstream consequences.

  • Magnitude scales with endurance volume and intensity, and with proximity in time to the lifting session. Low-volume, high-intensity intervals interfere far less than long steady-state volume does.
  • Separate conflicting sessions by ≥6 hours where possible, and put the priority quality first in the day when the nervous system is fresh.
  • Running interferes with lower-body hypertrophy substantially more than cycling does — the eccentric loading of running adds a damage cost that cycling largely avoids.
  • Practical resolution: one primary quality per 8–16 week block; hold the others at a maintenance dose. Maintenance is remarkably cheap — roughly a third of the volume that built a quality will hold it.

Sprint speed and vertical jump are compatible with each other and with strength work — they are all rate-of-force-development qualities. Marathon endurance conflicts with all three. Mobility conflicts with nothing and can run alongside anything.

Program one

Sprint speed

A horizontal-force problem that most people train as a leg-strength problem.

RCT / meta-analysisResearcher, own field

The mechanics — Morin and Samozino

Mechanism

Force-velocity profiling, and the orientation insight

Jean-Benoît Morin and Pierre Samozino's model treats sprint acceleration as a linear force-velocity relationship, characterised by two intercepts: F₀, theoretical maximum horizontal force at zero velocity, and V₀, theoretical maximum velocity at zero force. Maximum power sits at the midpoint. The genuinely important finding is not the profile itself but the *orientation* result: what separates fast accelerators is not how much total force they put into the ground but the ratio of horizontal to resultant force — the ability to direct force backwards rather than downwards as velocity rises. Two athletes with identical squat strength can differ enormously in acceleration because one applies force in a more useful direction.

Mechanism

Top speed is a different problem entirely

Acceleration is horizontal force; maximum velocity is vertical force applied in a vanishingly short ground contact. At top speed a sprinter's foot is on the ground for roughly 0.08–0.10 seconds, and elite sprinters produce ground reaction forces of 4–5× bodyweight within that window. Ken Clark and Peter Weyand's work shows top speed is determined primarily by how much force can be applied in the first half of stance, not by stride length, stride frequency, or how far behind the body the leg swings. The trainable variables are therefore limb stiffness, rate of force development, and hip flexor/extensor angular velocity — not 'longer strides'.

The honest counter-argument

A 2023 IJSPP paper argues bluntly that force-velocity profiling for sprint running is a dead end, on the grounds that the profile is largely determined by maximum power and adds little beyond simple split times. Morin's group disputes this. The field-level takeaway survives either way: sprint fast, sprint with resistance in the acceleration range, and get strong — the disagreement is about how much diagnostic value the profile adds, not about the training.

The weekly structure
2 quality sprint sessions per week, minimum 48h apart, never on the same day as heavy lower-body lifting if avoidable. Full recovery between reps is not optional — 1 minute per 10 m sprinted is the working rule. A sprint session done tired is a conditioning session that trains slow.
Session A — acceleration (F₀)
Sled pushes and heavy sled marches at loads causing ~50% velocity decrement (this is the load that maximises horizontal power output, per Morin's group), hill sprints at 5–10°, and 10–30 m sprints from a three-point start. 6–10 total reps.
Session B — maximum velocity (V₀)
Flying sprints of 20–30 m with a 20–30 m build-in, at 95–100% effort. 4–6 reps with 4–6 minutes' rest. This is the session that cannot be faked when fatigued, and the one most people skip.
The lifting that supports it
Heavy hip-dominant work for F₀ — trap bar deadlift, hip thrust (the most horizontally-oriented loaded hip extension available), Bulgarian split squats. Plus low-volume, high-intent jumps and Olympic derivatives for rate of force development. Two sessions weekly, 2–4 sets of 3–5 reps at high intent, never to failure — grinding reps train the opposite of what sprinting needs.

Program two

Marathon and distance

VO₂max sets the ceiling. Almost nobody is limited by it.

RCT / meta-analysisResearcher, own field

The three variables, in order of what actually limits you

Mechanism

VO₂max, fractional utilisation, and economy

Marathon performance is the product of three terms. VO₂max is the ceiling on oxygen flux. Fractional utilisation is the percentage of that ceiling you can hold for the duration — governed by lactate threshold, which is where lactate production exceeds clearance capacity. Running economy is the oxygen cost of a given pace, determined by tendon elastic energy return, muscle fibre type, and movement efficiency. Elite marathoners are not distinguished primarily by VO₂max — many club runners have comparable values. They are distinguished by holding 80–88% of it for two hours, and by exceptional economy.

Mechanism

The lactate shuttle — why lactate is not the villain

George Brooks' lactate shuttle overturned the old picture. Lactate is not a waste product causing fatigue; it is a fuel and a signalling molecule. Fast-twitch fibres produce it; slow-twitch fibres and the heart take it up via MCT1 transporters, convert it back to pyruvate, and oxidise it. Training raises both MCT1 density and mitochondrial capacity, so the same absolute workload produces less blood lactate. Both terms move, and Brooks' own account includes both: greater mitochondrial density shifts more pyruvate toward oxidation so less lactate is produced, *and* raised MCT1 clears more of what is produced. The popular framing that it is purely a clearance improvement is a simplification. Iñigo San Millán's Zone 2 framing is built on exactly this: the intensity that maximally stresses the clearance machinery without overwhelming it.

Mechanism

Seiler's 80/20 — an observation before it was a prescription

Stephen Seiler's polarised model came from measuring what elite endurance athletes across cycling, rowing, running and cross-country skiing actually do: roughly 80% of sessions at low intensity (below the first lactate threshold), ~20% high, and strikingly little in the middle. The mechanistic argument for avoiding the middle: threshold work accumulates autonomic and glycogen-depletion fatigue at close to the cost of high-intensity work while delivering a weaker mitochondrial and VO₂max stimulus. It is the worst return on fatigue of any zone.

The correction that matters for a longevity reader

80/20 is a distribution of *sessions*, not a licence to only do Zone 2. Mitochondrial and VO₂max adaptations are essentially confined to work above ~65% of peak power (Granata et al. 2018 meta-analysis; 2024 Calgary study). The 20% is not optional garnish — it is where the ceiling moves. Zone 2 builds the clearance capacity and the recoverable volume that let you do the 20% well.

Base phase (weeks 1–8)
80% easy volume, conversational, nasal-breathing-possible pace. 2 quality sessions weekly: one threshold (3–4 × 8–10 min at ~lactate threshold, 2 min float) and one VO₂max (4×4 min at 90–95% HRmax, or 30/15s). One long run building 10% per week with a down week every fourth.
Build phase (weeks 9–16)
Long run extends toward 2.5–3 h. Add marathon-pace segments inside the long run — this trains fractional utilisation specifically, which is the variable most likely to be limiting. Keep one true VO₂max session weekly to defend the ceiling.
The strength work almost every runner skips
2 sessions weekly of heavy, low-volume lifting improves running economy by 2–8% in meta-analysis, with no hypertrophy required. The mechanism is neural and tendinous, not muscular: increased tendon stiffness improves elastic energy return per stride, and improved rate of force development shortens ground contact. Heavy (≥85% 1RM), low reps, high intent. Plus plyometrics for the same reason.

Program three

Vertical jump

Force you can produce in 200 milliseconds — not force you can produce at all.

RCT / meta-analysisPrimary literature

Rate of force development and the stretch-shortening cycle

Mechanism

Why maximum strength stops predicting jump height

A countermovement jump takes roughly 200–300 ms of propulsion; a depth jump ground contact can be under 200 ms. Maximum voluntary force takes ~300 ms or more to develop. So beyond a moderate strength base — roughly a 1.5–2× bodyweight squat — additional maximum strength stops transferring, because you no longer have time to express it. The trainable quality becomes rate of force development: how steep the force-time curve is in its first 100–150 ms, which is governed by motor unit discharge rate and the speed of initial recruitment rather than by cross-sectional area.

Mechanism

The stretch-shortening cycle — three contributing mechanisms

A countermovement jump beats a static jump by 10–20%, and three mechanisms contribute. Elastic energy storage: the tendon stretches during the eccentric phase and recoils, returning energy that cost no ATP. Stretch reflex: rapid muscle lengthening activates muscle spindles, whose Ia afferents drive a monosynaptic reflex increasing motor unit activation. And pre-activation with active-state development: the muscle is already generating force before ground contact, so crossbridges are formed and the series-elastic component is taut when loading begins — the largest of the three contributions and the one most people ignore.

Mechanism

Tendon stiffness is the trainable substrate

Get the physics right, because the popular version is wrong: energy lost in a tendon is *hysteresis*, not compliance — the two are independent properties, and human tendon returns roughly 90–93% of stored energy regardless of how stiff it is. What stiffness governs is timing and transmission: a stiffer tendon transmits muscle force to the skeleton with less delay and less length change, which matters enormously when ground contact lasts under 200 ms. Stiffness adapts to habitual loading — plyometric training increases Achilles and patellar tendon stiffness measurably, even where maximal strength is unchanged, because tenocytes respond to strain magnitude and rate by upregulating collagen synthesis and cross-linking. This is why plyometric training improves the Reactive Strength Index (jump height ÷ ground contact time) more than heavy lifting does, despite lifting producing greater force.

Phase 1 — build the strength base if it isn't there
If back squat is below ~1.5× bodyweight, that is the limiting factor and heavy lifting is the highest-return work. 3–5 sets of 3–5 reps, squat and hip hinge, twice weekly.
Phase 2 — rate of force development
Once the base exists, shift emphasis to intent. Speed squats at 30–50% 1RM moved maximally fast, jump squats at 0–30%, trap bar jumps, and Olympic derivatives (high pulls, mid-thigh pulls — the pull is where the power is; the catch is a skill tax). 3–5 sets of 3, full recovery, terminate the session the moment bar speed drops.
Phase 3 — plyometrics, progressed by contact time
Progress by intensity of ground contact, not by volume. Pogo hops and ankle stiffness work → countermovement jumps → box jumps for landing mechanics → depth jumps from 30 cm, progressing height only while RSI improves. 60–120 foot contacts per session, twice weekly, on fresh legs. Depth jumps from too high a box lengthen ground contact and train the opposite quality.
Measure it or you are guessing
Track countermovement jump height and Reactive Strength Index (a phone slow-motion camera gets both). RSI below 1.5 means the stretch-shortening cycle is the limiter and plyometrics are the priority; RSI above 2.5 with a weak squat means strength is the limiter. The ratio tells you which phase you belong in.

Source tier — Primary literature throughout — plyometric/RSI meta-analyses (Sports Medicine, 2023), tendon adaptation and depth-jump muscle-tendon behaviour studies. Practitioner tier: Cal Dietz (triphasic training), Chris Korfist, and Frans Bosch — all with real coaching records and genuinely useful frameworks, none of which is a citation.

Program four

Mobility and flexibility

Mostly a nervous-system permission problem, not a tissue-length problem.

RCT / meta-analysisPrimary literature

What actually changes when you get more flexible

Mechanism

Stretch tolerance, not muscle lengthening

This is the finding that reorganises the whole topic. Short-to-medium-term flexibility gains are driven predominantly by increased *stretch tolerance* — a change in the sensory and nociceptive response to a given joint angle — rather than by any measurable change in resting muscle length. Studies measuring muscle-tendon unit stiffness before and after several weeks of static stretching find range of motion increases substantially while passive stiffness barely moves. Your muscle did not get longer. Your nervous system stopped defending that position. Note what this does *not* mean: Magnusson's work showed that during a slow static stretch the muscle is essentially electrically silent, so this is not a spinal reflex being overcome. The change is in the perception of and tolerance for the sensation — cortical and nociceptive rather than reflexive — which is precisely why flexibility is state-dependent, worse when anxious, and better when warm.

Mechanism

The exception — loaded stretching does change tissue

Longer-term structural change is real but requires load, not just duration. Training a muscle under tension at long muscle lengths adds sarcomeres in series, lengthening fascicles — the same sarcomerogenesis mechanism described in the hypertrophy literature, where titin's strain-dependent unfolding acts as the mechanosensor. This is why loaded end-range work (Jefferson curls, ATG split squats, deep-position eccentrics) produces more durable range than passive stretching, and why the strongest position in a new range is what makes the range keepable.

Mechanism

Why PNF works — and why the standard explanation is wrong

Contract-relax methods do produce larger acute range gains than passive stretching. The textbook explanation — autogenic inhibition, where Golgi tendon organ firing during a maximal contraction reflexively reduces motor neuron excitability in the same muscle — has been specifically tested and is not supported (Chalmers, Sports Biomechanics, 2004): EMG during the stretch phase does not show the predicted inhibition, and any reflex effect decays far faster than the range gain lasts. The better-supported account is the same one that explains stretching generally — raised stretch tolerance, plus mechanical creep in the tissue during the hold. The effect is largely acute: it opens a window in which to train the new range, and the range only persists if you produce force in it.

Assess before you program
The most common error is stretching everything. Test specific positions, find the two or three that are actually limiting, and train only those. A programme with 3–5 targeted stretches beats a follow-along routine that covers everything shallowly — this is exactly the model Matthew Smith teaches in his mobility toolkit, and it is the right one: individualised assessment first, minimal effective selection second.
Frequency, and the counterintuitive part
1–2 dedicated sessions per week per target position is sufficient, and more is often worse. Stretching produces genuine fatigue and the adaptation happens during recovery from it — the same logic that governs lifting volume. Daily aggressive stretching of the same position is the flexibility equivalent of junk volume.
The session
Warm first (tissue temperature genuinely changes compliance). Then per position: 2–3 sets of 60–120 s total, either long static holds, or contract-relax cycles (6 s contraction at ~50% effort, then deepen), or loaded end-range work. Finish every session with an *active* attempt to reach the new range unassisted — if you cannot get there under your own control, the range will not stay.
Timing around lifting
Long static stretching immediately before heavy or explosive work transiently reduces force output and rate of force development — a real effect, though smaller than commonly claimed and largely gone within 10 minutes. Put static work after training or in its own session; put dynamic work before.

Source tier — Primary literature for the stretch-tolerance mechanism (well-replicated across multiple groups). Operator tier: Matthew Smith — strength and mobility coach to Olympic athletes including swimmer Cameron McEvoy, trained under Charles Poliquin, and someone who learned adult flexibility from a genuinely inflexible starting point, which is the relevant credential. His assessment-first, minimal-selection, low-frequency model matches what the literature supports better than most of this space does.

Research notes, not medical or financial advice. Every prescription or experimental item named here is named with its mechanism and its risk and without a dose, on purpose — several require physician supervision, several are unregulated, and several are inappropriate for a body that is still developing. Start from your own bloodwork and a doctor, never from someone else’s regimen.

Non invenitur. Fit.