Strength standards

From 1.8 million judged lifts to one percentile

1. Take only judged lifts. Every row is an attempt performed on a competition platform in front of three referees. Anything from a phone, a survey or a training log is not in the archive at all.
2. Drop equipped lifts. Raw only — no squat suits, no bench shirts. A suit is worth tens of kilos and would silently inflate every figure on the page.
3. Count each lifter once, at their best. Somebody who competes twenty times is one data point, not twenty, so the curve is a population of people rather than a population of meets.
4. Split by sex and 5 kg bodyweight class, then by age band. A lift is a different achievement at 60 kg and at 120 kg, so a single national average would be an average of two different questions.
5. Sort the class and read off the 99 cut points. Your percentile is the position of your lift in that sorted list — not a score, a rank against named people.
6. Publish nothing thin. A slice with fewer than 500 lifters is not given a page: its tails would be noise dressed as a standard.

percentile(x)  =  100 × count(best_raw ≤ x) ÷ n

  class   =   sex  ×  5 kg bodyweight bucket  [ ×  age band ]
  n       =   lifters in that class, each counted once, at their best
  x       =   your lift, in kilograms

  band    =   01–19  novice        50–79  advanced      95–99  world class
              20–49  intermediate  80–94  elite

The full method, including what is excluded and why →