LiftGauge formulas, data and sources

§ 0 how to read this app

/calculator

The live 1RM calculator. Type a top-set (your heaviest set today) and read out estimated 1RM, plate-load, warmup ramp, and Prilepin volume — all from one screen.

/scores

Federation strength scores: DOTS, Wilks 2020, and IPF GoodLift computed in parallel from competition total + bodyweight + sex.

/rankings

Where your lift sits in OpenPowerlifting meet-result cohorts. The result names the selected axes, displays the meet-result entry count, and discloses any widening used. It does not estimate a general-population percentile.

/nutrition

Energy and macros from goal (cut/maintain/bulk), activity level, and body data. The PLAN tab shows kcal + protein/carbs/fat; STRENGTH shows the per-lift retention curve over 12 weeks.

/method

This page. Formula provenance, citations, and what the abbreviations mean.

/dataset

Downloadable artifacts with source-dump and latest-meet dates, sample sizes, caveats, citations, and rights stated per source. The OpenPowerlifting-derived cohort cube is CC0; that label is not applied blanketly to unrelated source material.

headline 1RM ≈ X kg ± Y %

The mean of seven 1RM formulas, rounded. ± is the spread between the highest and lowest formula — wider band, less certainty.

RPE

Rate of Perceived Exertion. 6 = easy, 10 = max effort. RPE 9 = "one rep left in the tank".

percentile

Rank inside the displayed OpenPowerlifting cohort. 70th percentile = higher than 70 % of qualifying lift observations in that cell — not 70 % of all lifters or all humans. (No "all humans" mode exists, see /method § 5.9.)

Other abbreviations. NPL = number of prescribed lifts per session. BW = bodyweight (kg). top-set = the heaviest working set of a session. PR = personal record. raw = belt, chalk and non-supportive sleeves only; Wraps, bench shirts and squat suits are excluded.

frequently asked

Six questions LiftGauge answers with cited sources, not opinion. Long-form provenance follows in §1–§8.

What is DOTS, and how does it relate to Wilks and IPF GoodLift?

DOTS is a strength-comparison coefficient by Konertz / BVDK (2019). It is the de-facto raw-open standard on OpenPowerlifting and on US raw federations such as USPA — but it is not an IPF formula. The IPF's own official coefficient has been IPF GoodLift (IPF GL Points) since 2020, when the IPF's formula review retired Wilks for systematically under-scoring heavier lifters (IPF formula evaluation, 2020). LiftGauge computes DOTS, Wilks 2020, and IPF GoodLift in parallel on /scores so you can see how the same total ranks under each system.

How accurate is Epley vs Brzycki for one-rep max?

Both are linear-ish approximations and disagree most at higher reps. At 5 reps the Epley estimate is typically within ~3 % of the actual 1RM for trained lifters; Brzycki tends to underestimate slightly past 6 reps. Validation: LeSuer et al. 1997 (J Strength Cond Res 11:211–213) tested seven 1RM equations across squat, bench, deadlift. LiftGauge shows all seven formulas (Epley, Brzycki, Lombardi, O'Conner, Wathan, Lander, Mayhew) with their spread surfaced — single-formula calculators hide the disagreement.

What dataset does LiftGauge's percentile rank use?

OpenPowerlifting public meet data (CC0 source dump dated 2026-05-16; latest included meet 2026-05-10). The trailing window beginning 2025-05-10 contains 186,303 eligible source rows; 186,218 rows contribute 615,721 qualifying lift observations, segmented by sex, lift, IPF bodyweight class, age (per year, 15–80), and equipment. A competitor may contribute multiple observations; n is not a unique-person count. Schema: Name · Sex · Equipment · BodyweightKg · Age · Best3SquatKg · Best3BenchKg · Best3DeadliftKg · TotalKg · Date · Place. Wraps, DQ and DD are excluded; approximate n.5 ages enter all-ages cells only. Sparse cells try exact age → all ages → ±1 / ±2 bodyweight classes, require n ≥ 30, and otherwise return no percentile; there is no global fallback. Source: openpowerlifting.org.

Does a rankings percentile represent all lifters?

No. The denominator is qualifying OpenPowerlifting lift observations in the displayed sex, lift, bodyweight-class, age, and equipment cohort. It excludes people who have never entered a recorded meet, and one person can contribute more than one observation. LiftGauge therefore reports the cohort and observation count instead of describing the percentile as population prevalence.

Why show seven 1RM formulas instead of one?

Because they disagree, and which one fits best depends on the lift, lifter, and rep range. With the implemented equations the full low-to-high range is about 5.6 % at 5 reps, 6.2 % at 8 reps, and 11.8 % at 12 reps. The mean is the headline; the displayed ±% is half that range, not a confidence interval.

What is FFMI and what is the natural ceiling?

FFMI (Fat-Free Mass Index) = leanKg / height_m² + 6.1 × max(0, 1.8 − height_m). Class thresholds (untrained <18, ..., elite <26, exceptional ≥26) come from Kouri et al. 1995 (Clin J Sport Med 5:223–228, PMID 7496846), which identified a "natural ceiling" around 25 based on drug-tested lifters. The 1.8 m height correction normalises against a reference individual. LiftGauge displays FFMI alongside BMI-derived BF% and waist/height to triangulate body composition, since each metric misclassifies different populations.

§ 1 1RM formulas

Full formula range ≈5.6 % at 5 reps and ≈11.8 % at 12. Trust the visible range, not a point estimate.

Legend: w = load lifted (kg) · r = reps performed · e = Euler's number (≈ 2.718). 1@10 anchor: at one rep, RPE 10, w = 1RM.

1.1Epley

1RM = w · (1 + r/30)

1.2Brzycki

1RM = w · 36 / (37 − r)

1.3Lombardi

1RM = w · r^0.10

1.4O'Conner

1RM = w · (1 + 0.025 · r)

1.5Wathan

1RM = 100 · w / (48.8 + 53.8 · e^{−0.075 · r})

1.6Lander

1RM = 100 · w / (101.3 − 2.67123 · r)

1.7Mayhew

1RM = 100 · w / (52.2 + 41.9 · e^{−0.055 · r})

RPE → %1RM source: Tuchscherer, M. Reactive Training Manual (RTS Press, 2008). RPE-load chart reproduced on /calculator. Anchor: 1@10 = 100 %.

§ 2 Prilepin's table

The active-row stamp on the table marks the zone matching your current /calculator input intensity. NPL = number of prescribed lifts per session in that zone; range is the productive band.

Medvedyev, A.S. A System of Multi-Year Training in Weightlifting. Trans. Andrew Charniga Jr. Sportivny Press, 1986. § 3.2.

intensity (% of 1RM)	reps/set	optimal NPL	range
55–65 %	3–6	24	18–30
70–80 %	3–6	18	12–24
80–90 %	2–4	15	10–20
≥ 90 %	1–2	7	4–10

§ 3 federation coefficients

Three formulas, three different reference populations — which is why the same total produces three different ranks.

3.1DOTS

Konertz / BVDK 2019. The de-facto raw-open standard on OpenPowerlifting and US raw federations (USPA) — not an IPF formula. Widely adopted because the older Wilks formula systematically under-scored heavier lifters.

3.2Wilks 2020

Wilks (1994), 2020-revision coefficients. Superseded at the IPF by IPF GoodLift in 2020, and on most raw scoreboards by DOTS. Kept for parity with old write-ups.

3.3IPF GL

IPF GoodLift 2020 — the IPF's official scoring coefficient, replacing IPF Points (2019) and Wilks. Raw and equipped tables differ; we expose raw open here.

§ 4 data attribution

openpowerlifting.org · CC0 · source dump 2026-05-16 · latest included meet 2026-05-10. We do not show beginner / intermediate / advanced / elite badges. Use percentile or don't.

§ 5 cohort data — /rankings

The /rankings view aggregates OpenPowerlifting meet records into a versioned, pre-computed cohort cube. The runtime never queries OpenPowerlifting — the cube is fetched from LiftGauge and stored in a dedicated runtime cache after first load.

5.1Source

openpowerlifting.org public dataset. License: CC0. Schema: Name · Sex · Equipment · BodyweightKg · Age · Best3SquatKg · Best3BenchKg · Best3DeadliftKg · TotalKg · Date · Place.

5.2Build

—

5.3Release window

—

5.4Cohort axes

Sex (M/F) × Lift (squat/bench/deadlift/total) × BW class (IPF: M 53/59/66/74/83/93/105/120/120+; F 43/47/52/57/63/69/76/84/84+) × Age (individual years 15–80, or all-ages) × Equipment (raw / single-ply / multi-ply). Wraps-labelled rows are excluded. Leave age blank for an age-agnostic comparison.

5.5Trajectory

Built from full meet histories (not limited to the 12-month window), grouped on Name + Sex + rounded 10 kg bodyweight bucket and the same resolved age cohort used by the lead percentile. Approximate OpenPowerlifting n.5 ages do not enter exact one-year bins. These are bodyweight-stabilised trajectory identity keys, not verified people. Cells with n < 30 are suppressed.

5.6Sparsity

Cohort widening hierarchy: exact year → all ages → ±1 bodyweight class → ±2 bodyweight classes. Each candidate must contain at least 30 qualifying lift observations. Widening is shown explicitly in the source row; if no candidate reaches n ≥ 30, the view returns no percentile instead of substituting a global cohort.

5.7Bias

IPF / USAPL meets are over-represented (more reporting). Retired lifters' last-meet values are frozen — their "current" is whenever they stopped competing. Equipped → raw is never auto-converted (different sport-physics, different cohort).

5.8Comparison population

OpenPowerlifting meet-result observations only. Of 186,303 source rows passing the 12-month source filters, 186,218 contribute at least one valid lift and produce 615,721 squat, bench, deadlift or total observations. The data spans drug-tested and untested federations. One person can contribute multiple observations; the displayed n is an observation count, not a unique-person count.

5.9Civilian baseline — why we don't ship one

No nationally representative dataset tests barbell squat, bench, and deadlift 1RM directly on civilian adults. Mass fitness surveys avoid barbell 1RM testing because it needs spotters and racks and introduces injury risk. /rankings therefore has no all-adults or general-population mode.

5.10What we don't show

No predicted next-PR. No "elite / intermediate / beginner" badges. No vanity-thresholds. The numbers are the cohort distribution; interpretation is yours.

§ 6 what we don't do

No accounts. No ads. No newsletter. No "Pro" tier. Numbers stay on your device. If localStorage is unavailable (private browsing), the calculator still works — you just lose persisted preferences.

§ 7 input keyboard controls

↹↑ / ↓

bump load by 2.5 kg / 5 lb (when load-input focused)

§ 8 nutrition model — TDEE, macros, retention

The /nutrition view uses published nutrition science for the base formulas but combines them in a custom-developed strength-retention model. This section explains what is published and what is liftgauge-proprietary.

8.1Mifflin-St Jeor BMR

BMR (M) = 10·kg + 6.25·cm − 5·age + 5; (F) = same but −161 instead of +5. Mifflin et al. 1990, Am J Clin Nutr 51:241–247 (PMID 2305711).

8.2Activity factors (TDEE multiplier)

1.20 sedentary / 1.375 lightly active / 1.55 moderate / 1.725 active / 1.9 very active. Standard ACSM Harris-Benedict revision (ACSM Guidelines for Exercise Testing and Prescription, 11e 2022, Ch.7).

8.3Protein target

Cut: 2.2 g/kg. Bulk/maintain: 1.8 g/kg. Helms et al. 2014 (J Int Soc Sports Nutr 11:20, PMID 24864135) + Morton et al. 2018 (Br J Sports Med 52:376–384, PMID 28698222) — meta-analysis on protein targets for resistance-trained adults.

8.4Fat + carbohydrate split

Fat: 25 % cut / 28 % maintain / 30 % bulk of target kcal. Carbohydrates = remainder. Within ACSM's 20–35 % fat-of-energy recommendation. ACSM GETP 11e + ISSN Position Stand Aragon et al. 2017 (J Int Soc Sports Nutr 14:16, PMID 28630601).

8.5Cut/bulk thermodynamics

Daily deficit/surplus converted via 7700 kcal/kg of fat tissue. Wishnofsky 1958 constant — accepted first-order approximation, modified by Hall et al. 2011 (Lancet 378:826), which shows the linear model underestimates the plateau after ~6 months. The plateau is approximated inside the projection via a rolling adaptation factor (max 18 %) rather than a separate adaptive-TDEE model, because the view does not log weekly calorie or weight series.

8.6BMI-based body-fat estimate

BF% = 1.20·BMI + 0.23·age − 10.8·sex − 5.4 (sex: M=1, F=0). Published population formula from Deurenberg, Weststrate & Seidell 1991 (Br J Nutr 65:105–114, PMID 2043597). Can misclassify muscular lifters; FFMI and waist/height are shown alongside.

8.7FFMI — Fat-Free Mass Index

FFMI = leanKg/(height_m²) + 6.1·max(0, 1.8 − height_m). The height correction normalises against a 1.80 m reference individual. Class thresholds (untrained <18, ..., elite <26, exceptional ≥26) derived from Kouri et al. 1995 (Clin J Sport Med 5:223–228, PMID 7496846), which identified a "natural ceiling" around 25 based on drug-tested lifters.

8.8Strength retention model (proprietary)

weeklyRetentionRate computes the change in weekly 1RM preservation as the sum of energy-status + protein + bodyfat + cut-duration terms. The bulk side follows Helms et al. 2023: a larger (15 %) vs smaller (5 %) surplus produced similar squat strength and muscle-thickness gains while the extra bodyweight tracked skinfold/fat gain — so 5–20 % surpluses share a similar strength term and larger surpluses mainly raise fat risk. The cut side is derived from Helms 2014, Garthe 2011, and Murphy & Koehler 2022. Coefficient weights and refusal thresholds are liftgauge-derived safety margins, not published prognostic models. Use retention as direction, not as an absolute number.

Summary: Base formulas (Mifflin-St Jeor, Deurenberg BF%, Kouri FFMI, Wishnofsky thermodynamics) are published and freely verifiable. The retention model + refusal thresholds are liftgauge-derived indicators built on top of published thresholds, not absolute predictions.