Used directly. The metric would not exist without this source.
Receipts
Every paper.
Every paper.
Every assumption.
The literature, frameworks, and platform sources our methodology leans on — listed plainly, with our adaptations called out where we adapted, and our assumptions called out where we made them up.
We don't sell certainty. We sell a methodology that shows its work.
Four tags for source type.
Not every reference plays the same role. We label each one so you can tell foundational research from a framework we admire from an Apple framework we use as-is.
Referenced but not the default. Available as a user option.
Used conceptually. Informs how we think, not the exact formula.
An Apple-shipped framework. Used as provided.
Organized by what it underpins.
Click into the muscle-nerds page for the deep-dive. Use this list when you want to cite us, replace one of our choices, or just see the bibliography.
Strength estimation
2 sourcesEpley, B. (1985). Boyd Epley Workout.
Brzycki, M. (1993). Strength testing — predicting a one-rep max from reps-to-fatigue.
Effort scaling
2 sourcesZourdos, M. C. et al. (2016). Novel resistance training-specific RPE scale measuring repetitions in reserve. J Strength Cond Res.
Helms, E. R. et al. (2018). RPE-based programming for resistance training. J Hum Kinet.
Hypertrophy volume
2 sourcesSchoenfeld, B. J. et al. (2017). Dose-response relationship between weekly resistance training volume and muscle hypertrophy. Sports Med.
Schoenfeld, B. J. et al. (2019). Resistance training volume enhances muscle hypertrophy but not strength in trained men. J Sports Sci.
Programming framework
2 sourcesIsraetel, M. (Renaissance Periodization). MEV / MAV / MRV landmark framework.
Helms, E. R. (The Muscle and Strength Pyramid).
Movement taxonomy
1 sourceCook, G. (Movement-pattern framework, abbreviated for practical lifting).
Recovery / rest
1 sourcede Salles, B. F. et al. (2009). Rest interval between sets in strength training. Sports Med.
Tempo
1 sourceWilson, J. M. et al. (2014). Effects of tempo on muscle hypertrophy and strength. J Strength Cond Res.
Statistical methods
1 sourceBox, G. E. P. (1976). Science and Statistics. J Am Stat Assoc.
Platform — Apple
4 sourcesApple — HealthKit (HKWorkout, HKQuantityTypeIdentifier).
Apple — Foundation Models framework.
Apple — ActivityKit (Live Activities) + Dynamic Island.
Apple — MusicKit + Apple Music metadata API.
The assumptions we made ourselves.
Every analytic in flexRep has at least one place where we had to choose. These are those places — severity-rated, documented, and open to disagreement.
Fractional-sets weighting is a chosen midpoint.
The literature establishes that secondary mover work contributes to hypertrophy at a lower rate than primary mover work. The specific weighting we use is a practical midpoint, not a published constant. Your disagreement with our weighting is really a disagreement with the meta-analysis design.
Stall detection windows are calibrated, not validated.
We chose a stall window long enough to filter noise and short enough to catch plateaus before they cost months. The thresholds are our own and have not been validated against a control population.
Movement-pattern inference is heuristic.
For curated exercises in the library, movement patterns are hand-tagged. For user-created and imported exercises, the pattern is inferred from name + primary muscle. Inference is high accuracy, not perfect.
e1RM is constrained to a sensible rep range.
Single-rep-max estimation formulas fit the working-rep range well and degrade beyond. We cap rep range used for e1RM rather than extrapolate beyond it.
All analytics compute on one lifter's data.
No normative population comparison. No "you are above average for your weight class." Your numbers are measured against your numbers. By design.
PR catalyst rankings are correlation, not causation.
When a song shows up in five PR sessions, we tell you. We do not tell you the song caused the PR. The leaderboard is labelled accordingly.
The strength glyph and rhythm waveform are decorative.
They are generative visualizations driven by your data. They are not diagnostic instruments. They look great in a share-card and that is the entire goal.
Disagreements welcome
If you'd score it differently, score it differently.
The same export that ships our analytics ships every set you've ever logged. The choices on this page are visible because we expect serious lifters to disagree with at least one of them — and we'd rather you swap our weighting for yours in a notebook than abandon the data because we hid the choice.
The methodology is a starting point. The data is yours. The math is whoever's it needs to be.
Read them. Argue with them. Use the data.
Every number in the app is grounded in something on this page. Every choice that wasn't grounded is labelled as such. No black boxes.