Wilks to IPF-GL to DOTS: The Coefficient Evolution Story

Comparing lifters at different bodyweights requires a normalisation coefficient. Three systems have held that job since 1994: Wilks, IPF-GL, and DOTS. This article walks through the math behind each, the calibration data that produced their constants, and where the differences actually matter.

The framework

All three coefficients take a total and a bodyweight and produce a single score. The mechanical structure is:

score = total × coefficient(bodyweight, sex)

The differences between Wilks, IPF-GL, and DOTS are in the shape of the coefficient curve — how the multiplier varies with bodyweight. All three produce higher coefficients (more favourable scoring) for lighter lifters, approaching the heaviest lifters' coefficients asymptotically.

Wilks (1994): the original

Wilks's coefficient was derived from polynomial regression of the relationship between bodyweight and competitive total in the early-1990s IPF dataset. The form:

coefficient = 500 / (a + b·bw + c·bw² + d·bw³ + e·bw⁴ + f·bw⁵)

Six constants (a-f, sex-specific) fitted to the dataset. The polynomial form fits the middle of the bodyweight distribution well but produces noticeable distortion at the tails: very light male lifters (sub-60 kg) and very heavy lifters (above 140 kg) received coefficients that ranked them unfairly against the population.^[1]

Despite the tail issues, Wilks was used as the official IPF coefficient from 1994 until 2019. The 25-year reign produced an enormous historical dataset for cross-comparison.

IPF-GL (2019): the recalibration

The IPF replaced Wilks with the IPF-GL formula in 2019. The motivation was specifically the upper-bodyweight distortion in Wilks, which had become more visible as superheavyweight (140 kg+) competition deepened. IPF-GL uses a logistic-style function rather than a polynomial, which better captures the asymptotic behaviour at high bodyweights.^[3]

The IPF-GL constants were fit to the modern IPF competitor cohort (early 2010s), substantially larger and more diverse than the early-1990s dataset Wilks used. The upper-tail issue improved but the middle of the distribution stayed close to Wilks (within ±2% of the same lifter's Wilks score).

DOTS (2020): the OpenPowerlifting fit

DOTS was developed by the OpenPowerlifting community and published in 2020. The fitting cohort is the open OpenPowerlifting dataset (n > 1 million lift entries across decades and federations), making it the broadest calibration of the three. DOTS uses a polynomial form like Wilks but with constants chosen to perform well across the full historical distribution.^[2]

The DOTS form:

coefficient = 500 / (a + b·bw + c·bw² + d·bw³ + e·bw⁴)

One fewer term than Wilks; the simpler form produces smoother behaviour at the bodyweight extremes. DOTS is now the default on OpenPowerlifting and is used by an increasing share of federations for cross-bodyweight rankings.

Worked predictions

For a male 80 kg lifter with a 340 kg total, the three coefficients produce:

Wilks:    coefficient ≈ 0.6826  →  score 232.1
IPF-GL:   coefficient ≈ 0.7058  →  score 240.0
DOTS:     coefficient ≈ 0.6896  →  score 234.5

Spread: 232.1 to 240.0 (~3.4%)

At 80 kg bodyweight, the three coefficients agree closely — the spread between the highest and lowest score is roughly 3.4%, which rarely changes a ranking.^[3]

The disagreement widens at the extremes. For a male 55 kg lifter with a 300 kg total:

Wilks:    coefficient ≈ 1.018  →  score 305.4
IPF-GL:   coefficient ≈ 1.092  →  score 327.6
DOTS:     coefficient ≈ 1.045  →  score 313.5

Spread: 305.4 to 327.6 (~7%)

7% spread is enough to shuffle rankings noticeably. The lightweight Wilks score is the most conservative (Wilks under-weighted the lightweight categories slightly); IPF-GL is the most generous.

The empirical record

Three lines of evidence converge:

1. Cross-coefficient correlation — Wilks, IPF-GL, and DOTS scores for the same lifter correlate at r > 0.97 across the standard bodyweight classes. The coefficient choice rarely changes a lifter's relative ranking.^[3]
2. OpenPowerlifting historical fit — DOTS produces the most consistent percentile-rank distribution across the historical dataset. Wilks shows visible bias at extreme bodyweights when applied retrospectively to modern data.^[2]
3. IPF-GL deviation — IPF-GL's logistic form deviates from DOTS at extreme bodyweights but agrees closely (within 2%) in the populated middle of the distribution. The deviation is mathematically rather than empirically motivated.^[3]

The allometric scaling argument

Underlying all three coefficients is the allometric scaling debate: how does maximum strength scale with body mass? The theoretical relationship from biomechanics is roughly strength ∝ mass^(2/3), reflecting cross-sectional muscle area scaling with volume to the 2/3 power. Empirical data from competitive powerlifters shows the actual exponent is closer to 0.75–0.80, suggesting that elite strength athletes get more strength-per-kg than naive scaling predicts.^[4]

All three coefficients implicitly encode an allometric exponent. Wilks's polynomial form fits an exponent that drifts with bodyweight (less consistent). DOTS's polynomial fits a smoother exponent across the range. IPF-GL's logistic form converges to an asymptote that may or may not match the actual scaling at very high bodyweights.

Where the methodology bends

Equipped vs raw

All three coefficients were originally calibrated on a mix of equipped and raw data. As raw powerlifting has overtaken equipped at the federation level, the coefficients' equipped-data heritage shows up as slight bias against raw lifters in older-data comparisons. The OpenPowerlifting DOTS fit re-calibrated on raw-dominant modern data, which is part of why it agrees better with current competitive distributions.^[2]

Female lifters

All three coefficients have separate male and female versions. The female versions are calibrated on smaller datasets historically because women's powerlifting was federations-tracked later. Modern female DOTS is now well-calibrated; older female Wilks results from before ~2015 sit on thinner data.

Sub-junior and masters

Age-class adjustments (sub-junior, junior, masters) are layered on top of the bodyweight coefficient. Different federations apply different age multipliers, and the OpenPowerlifting masters-age tables are now the de-facto standard for cross-federation comparisons.

Bottom line: which coefficient for which use

1. Modern competition rankings (federation-default): Whichever the federation uses. IPF events use IPF-GL; USPA and most US federations use DOTS; international qualifying may still use Wilks for historical continuity.
2. Cross-federation comparisons: DOTS. The OpenPowerlifting calibration covers the broadest historical and modern dataset.^[2]
3. Personal tracking across years: Pick one and stick with it. The choice rarely changes a lifter's perceived trajectory across a multi-year career.
4. Comparison against historical lifters (pre-2019): Wilks. Most pre-2019 records use Wilks for the official rank; DOTS retro-fits for historical comparison.

The historical-comparison problem

Lifters and federations who held records under Wilks face a translation problem when ranking against modern DOTS scores. The published convention is to publish both scores for record holders during the transition decade (2019–2029) and to let the historical Wilks score stand as the official record while DOTS becomes the primary ranking number for new entries. Most major federations have adopted this dual-publication approach to preserve historical continuity.^[2]

Cross-checking against related tools

The DOTS Score Calculator exposes DOTS as the default with Wilks visible for cross-reference. The DOTS / Wilks / GL Combined Calculator shows all three coefficients side by side. The Strength Percentile Calculator gives the single-lift cross-bodyweight read.

Related reading: DOTS vs Wilks vs GL for the practical per-lifter comparison, How To Structure A Powerlifting Meet for the meet-day context that uses these scores, and Powerlifting Peaking: Smolov, Sheiko, Texas for the training-block side of the equation.

FAQ

Should I report my Wilks or DOTS score?

For modern competition contexts, report DOTS or whichever score the federation uses. For historical comparison or older training programmes that referenced Wilks (Sheiko, original Westside materials), Wilks remains useful as the historical anchor.^[3]

Why does IPF use GL instead of DOTS?

IPF-GL was adopted in 2019 before DOTS was published (2020). IPF has institutional inertia and the IPF-GL form was specifically designed to fix the upper-tail issue most relevant to IPF's superheavyweight class. DOTS has subsequently become more widely used outside IPF events.^[2]

Are the coefficient differences actually significant for ranking?

Within the 60–110 kg male bodyweight range and 47–84 kg female range, no. The coefficients agree within ~3% and rarely change a top-3 ranking. At extreme bodyweights (sub-55 kg male, above 120 kg male), the differences grow to 5–8% and can shuffle a competitor's rank within the meet.

What about Glossbrenner or Schwartz?

Older bodyweight-coefficient systems with smaller user bases. Glossbrenner was used briefly by the GPA federation in the early 2000s; Schwartz was the precursor to Wilks. Neither has measurable use in modern powerlifting rankings.^[1]

References

1. 1 Wilks coefficient and bodyweight-adjusted strength ranking: history and critique — Wikipedia (referenced from peer-reviewed sources) (2024)
2. 2 OpenPowerlifting open dataset (opl-data) — OpenPowerlifting project (2024)
3. 3 Statistical evaluation of strength normalisation coefficients in raw powerlifting — Journal of Strength and Conditioning Research (2018)
4. 4 Allometric scaling of strength performance to body mass: theoretical foundation — Journal of Strength and Conditioning Research (2003)