Methodology

Pelland JC, Remmert JF, Robinson ZP, Hinson SR, Zourdos MC (2025). The Resistance Training Dose Response: Meta-Regressions Exploring the Effects of Weekly Volume and Frequency on Muscle Hypertrophy and Strength Gains. Sports Medicine 56(2):481-505. PMID 41343037

The entire volume model is this one curve, refitted from the authors’ own open dataset. It sets the smallest weekly dose that produces a measurable change, four sets a week, and it prices every set after that against the first one. That price is the number the app puts next to each muscle.

Bartolomei S, Sadres E, Church DD, et al. (2017). European Journal of Applied Physiology. PMID 28447186

Eight sets of ten at 70% left peak force impaired at 72 hours, while eight sets of three at 90% was not impaired at any timepoint, at the same set count. This is why the model charges reps at moderate load more than heavy low-rep work, and it reverses the common assumption that heavy days are the expensive ones.

Moran-Navarro R, Pérez CE, Mora-Rodríguez R, et al. (2017). PMID 28965198

Sets taken to failure slowed recovery of jump height, bar velocity and creatine kinase to 24 and 48 hours against a volume-matched non-failure protocol. Failure, not set count, is the variable the model prices.

González-Badillo JJ, Rodríguez-Rosell D, Sánchez-Medina L, et al. (2016). PMID 26667923

Jump height was still depressed 48 hours after sets to failure and recovered within six hours at half the reps. It sets the shape of the failure-proximity ramp.

Pareja-Blanco F, Villalba-Fernández A, Cornejo-Daza PJ, et al. (2019). PMID 30836680

More fatigue and slower recovery at the lower load with high velocity loss than at 80% with low velocity loss. Load alone does not predict recovery cost, which is why the model does not take %1RM as an input.

Thomas K, Brownstein CG, Dent J, et al. (2018). PMID 30067591

Ten sets of five at 80%, not to failure, depressed twitch force for 48 hours. We publish this one as a known limitation: our decay constant reads only proximity to failure, so it under-charges heavy high-volume work that stops short of failure.

Farrow J, Steele J, Behm DG, et al. (2021). PMID 32401690

Separates volume and proximity to failure as independent drivers of fatigue. That separation is the structure the model encodes.

Damas F, Phillips SM, Libardi CA, et al. (2016). PMID 27219125

The same lifter’s post-session force deficit went from 22% down in week one to 3% down in week three of the same program. Unfamiliar work is the single largest identified source of recovery variance, so the model carries a novelty multiplier that decays per muscle.

Newton MJ, Morgan GT, Sacco P, et al. (2008). PMID 18550979

Trained men recovered by day three from a bout that left untrained men 40% down on day three. Training history scales the novelty cost rather than resetting it.

Chen TC, Chen HL, Lin MJ, et al. (2011, 2023). PMID 20852880 · PMID 36427271

Chen 2011 tested muscle size as a predictor of recovery and rejected it, finding small arm muscles recovering more slowly than quads. That is why the engine carries no per-muscle recovery table and treats every muscle the same. Chen 2023 showed three weeks of disuse made the same muscle measurably more damageable, which is what the novelty cost on an unaccustomed movement is built from.

Hyldahl RD, Chen TC, Nosaka K (2017). PMID 27782911

The repeated-bout protective effect is induced by one exposure and most of the protection arrives early. That is why the novelty multiplier decays over a few sessions rather than many.

Nosaka K, Clarkson PM (1995). PMID 8531624

Repeated bouts on already-damaged muscle neither worsened the damage nor slowed recovery, so accumulated demand should saturate rather than stack forever. The model’s clamp is that saturation.

Marshall PWM, Cross R, Haynes M (2018) · Marshall PW, Metcalf E, Hagstrom AD, et al. (2020). PMID 28716692 · PMID 31343516

Full-body sessions in trained men left voluntary activation and EMG unchanged at 24 and 48 hours.

Behm DG, Alizadeh S, Hadjizadeh Anvar S, et al. (2021). PMID 33818751

Non-local muscle fatigue meta-analysis: a standardised mean difference of negative 0.02 across 303 participants. With Marshall, this is why nothing in the app models a systemic or central fatigue budget. It models local recovery of the muscles that did the work.

Grandou C, Wallace L, Impellizzeri FM, et al. (2020). Sports Medicine. PMID 31820373

Sustained performance decline is the only fatigue marker the overtraining literature validates. It leads our trigger list for that reason, and it is why the app does not ask you about heart rate variability or hormonal markers.

Coleman M, Burke R, Augustin F, et al. (2024). PeerJ. PMID 38274324

A one-week complete break mid-block did not improve hypertrophy and carried a small, uncertain strength cost. We never prescribe cessation, only a reduction.

Bosquet L, Montpetit J, Arvisais D, Mujika I (2007). Medicine & Science in Sports & Exercise. PMID 17762369

The optimal taper cuts volume 41 to 60% and leaves intensity and frequency alone. That is exactly the shape of a LiftingLabs deload week.

Travis SK, Mujika I, Gentles JA, et al. (2020). Sports. PMID 32917000

Volume cuts beyond 70%, or more than seven days off, cost strength. This sets the floor on how far a deload is allowed to reduce your work.

Pritchard HJ, Barnes MJ, Stewart RJ, et al. (2018). Journal of Strength and Conditioning Research. PMID 29369954

Two to five days of complete rest increases strength expression. Why a short break before a heavy attempt is not treated as lost training.

Rogerson D, Nolan D, Androulakis Korakakis P, et al. (2024). Sports Medicine Open. PMID 38499934

Coaches deload every 5.6 weeks on average for 6.4 days, and only 13% do it purely in reaction to fatigue. Our fallback trigger at seven weeks comes from this survey, not from a trial.

Refalo MC, Helms ER, Robinson ZP, Hamilton DL, Fyfe JJ (2024). Journal of Sports Sciences 42(1):85-101. PMID 38393985

Across fifteen studies, training closer to failure helped hypertrophy with effect sizes of roughly 0.15 to 0.21. Small and real, which is why accessory targets sit at one to three reps in reserve rather than at failure.

Robinson ZP, Pelland JC, Remmert JF, Refalo MC, Jukic I, Steele J, Zourdos MC (2024). Sports Medicine 54(9):2209-31. PMID 38970765

Proximity to failure modelled as a continuous predictor across 55 hypertrophy and 67 strength studies. The hypertrophy slopes are negative with confidence intervals excluding null, so growth keeps rising as sets end nearer failure, with no taper. The strength slopes contain null. Effort targets are set by fatigue cost, not by growth potential.

Bickel CS, Cross JM, Bamman MM (2011). Exercise dosing to retain resistance training adaptations in young and older adults. Medicine & Science in Sports & Exercise 43(7):1177-87. PMID 21131862

Weekly volume cut by 89%, from 27 sets to 3, with load held, and 1RM still rose across 32 weeks. The best-grounded number in the whole volume model is the maintenance floor, and this is why.

Barsuhn A, Wadhi T, Murphy A, et al. (2025). Journal of Applied Physiology. PMID 39665246

Adding 30% and 60% over a maintenance dose produced no detectable size difference. Used as a brake on the assumption that more is reliably better.

Enes A, Alves RC, Schoenfeld BJ, et al. (2021) · Enes A, De Souza EO, Souza-Junior TP (2024). PMID 34260860 · PMID 37796222

Volume-matched and higher-volume comparisons in resistance-trained lifters that did not separate. Part of why the target band is a range rather than a point.

Wendler J (2009). 5/3/1.

The training max convention, roughly 90% of a true 1RM, and the four-week microcycle. Used across every program in the engine.

Tuchscherer M (2009). The Reactive Training Manual.

The reps-in-reserve to percentage chart the engine uses to back-solve a day-of 1RM estimate from your reported RPE and re-prescribe the rest of the session against it.

Epley B (1985) · Brzycki M (1993).

Closed-form 1RM estimation for sets logged without an RPE. Blended with the RPE-derived estimate at 0.7 to 0.3 when both exist.