About Strength Level

2007: Original idea and exact-bodyweight strength ratings

Michael later recalled conceiving Strength Level over lunch at college and coding the first version that weekend. He publicly announced the calculator on 6 June 2007 because neither absolute lift weight nor a simple lift-to-bodyweight ratio gave him a fair way to judge progress across different body sizes. The calculator rated bench press, squat, deadlift, power clean, and clean and press for male and female lifters at the exact bodyweight entered. Compare the current calculator with the original launch post and archived 2007 calculator.

Michael began with strength standards provided by Lon Kilgore, PhD, and published by ExRx. ExRx stated that the tables drew on definitions in Practical Programming by Kilgore, Mark Rippetoe and Glenn Pendlay; techniques from Starting Strength by Rippetoe and Kilgore; the authors' experience and judgement; and published performance standards for powerlifting and weightlifting.

The source tables organised performance into fixed bodyweight rows and expressed lift weights in pounds. Michael used regression to transform them into continuous formulas. The resulting implementation stored four coefficients for each exercise and gender in a continuous two-variable model, including a bodyweight-by-level interaction. At a fixed level, required lift weight changed linearly with bodyweight. At a fixed bodyweight, the calculator inverted the model to return a decimal rating in kg or lb, including ratings between levels on the original Untrained (1) to Elite (5) scale. Strength Level's rounded kg and lb tables used practical bodyweight intervals and were generated from the model rather than used as calculator lookup rows. Michael later documented the method in a 2015 account of Strength Level's early years.

On 19 June 2007, Michael also described using the separate exercise ratings to identify weak points and encourage balanced strength across the main lifts.

2015: Community evidence and percentile innovation

By mid-2015, lifters could enter working sets of one to ten repetitions. We converted each set to an estimated one-rep maximum (1RM) and used those results to build standards for different bodyweights. As more results arrived, we updated the standards and began showing each exercise's supporting lift count publicly. This created a continuing loop from real lifter input to updated standards and visible evidence.

On 7 March 2015, Strength Level launched an overall percentile across multiple lifts. The first version compared each lift with the performance expected at that bodyweight, then combined them into one score: "Overall you are stronger than X% of users."

In July 2015, we replaced the original formulas derived from Kilgore's standards with new regression models fitted to more than 500,000 community lifts. Compare the current male strength standards with the archived 2015 version, and the current standards overview with the superseded original tables.

Michael defined five thresholds: Beginner means stronger than 5% of lifters, Novice 20%, Intermediate 50%, Advanced 80%, and Elite 95%. Symmetric around the median, they translated the statistical model into a clear five-level scale. The original calculator had followed the Untrained-to-Elite training-status labels in Kilgore's standards, published by ExRx. Strength Level replaced Untrained with Beginner and explicitly defined every level by percentile. This is the earliest documented use we have found of that exact Beginner-to-Elite percentile ladder for strength standards. The format is now common across strength websites.

In October 2015, the original overall score was replaced by a dedicated male and female Big Three ranking for squat, bench press and deadlift. A November 2015 explanation of overall strength showed how the total was converted into a bodyweight-specific percentile and displayed alongside star ratings for the individual lifts.

2016: Data-driven exact-bodyweight and personalised standards

In May 2016, we extended the community-data standards by jointly modelling lift weight against bodyweight. This brought the original calculator's exact-bodyweight comparison into the data-driven standards system without relying on fixed bodyweight classes.

Lift analysis: We launched a lift analysis that explained the full calculation: entered repetitions, estimated 1RM, age-adjusted exact-bodyweight percentile, Beginner-to-Elite rating, and simple bodyweight ratio.

Age: We introduced age-adjusted standards. The current model uses ages 25-40 as its adult baseline and adjusts the standards for lifters younger than 25 or older than 40.

Bodyweight exercises: We added repetition standards for Pull Ups and Dips, including bodyweight-only, weighted, and assisted versions. From 2022, public tables showed standards as either repetitions or 1RM added/assistance weight, with both views shown by bodyweight and age.

2017: Standards across the gym

Dumbbell, machine, and cable standards joined barbell and bodyweight exercises. Early additions included Dumbbell Bench Press, Dumbbell Row, Dumbbell Curl, Dumbbell Shoulder Press, Sled Leg Press, Horizontal Leg Press, and Lat Pulldown. We also launched direct exercise-to-exercise standards comparisons.

2018: Powerlifting total percentiles

We launched a calculator that ranked the combined Squat, Bench Press, and Deadlift total for the lifter's exact bodyweight and gender, alongside Wilks. Dedicated powerlifting standards followed in 2019.

2020: Running Level and a common fitness percentile

We launched Running Level, extending Strength Level's percentile framework beyond lifting for the first time. It converted race times into age-adjusted percentiles and used the same 5%, 20%, 50%, 80% and 95% Beginner-to-Elite levels. Standards for distances including 5K, 10K, Half Marathon and Marathon showed that strength and endurance could be understood through one consistent system.

2021: One percentile system across five sports

We expanded the framework to Rowing Level, Cycling Level, and Swimming Level. Their models handled indoor-rowing times, cycling distance, power and bodyweight, and swimming times while returning the same percentile and Beginner-to-Elite level. By the end of 2021, all five sports shared one performance language.

2026: Data-backed standards for every exercise

The first full standards refresh since 2022 covers 287 exercises using 195,513,376 lifts from 27,893,268 users. Every exercise now has male and female standards built with direct data from that exercise. When we compared equivalent old and new models on the same data, 356 of 368 new models (97%) had lower prediction error. Direct-data coverage improved most for female standards, including Hip Thrust, Horizontal Leg Press and Leg Extension. For 17 female exercises with limited direct data, we also used evidence from closely related exercises to improve reliability. Explore the 2026 standards.