Enter your bodyweight and lifts.
Who you're compared with: every lifter in OpenPowerlifting who competed raw (no wraps, suits or shirts) in a drug-tested full-power meet, aged 16+ (or age unknown), not disqualified. Each lifter counts once, at their best ever for that lift, with the bodyweight from that meet: 230,000 men and 110,000 women.
Your percentile is the share of those lifters whose best is below yours, with lifters near your bodyweight counting most:
P(x, b) = Σ wᵢ · [liftᵢ < x] / Σ wᵢ
wᵢ = exp(−½ · ((ln bwᵢ − ln b) / 0.035)²)
The weight is a bell curve on log bodyweight, about ±3.5% wide, so "near" scales with you (±2 kg at 60 kg, ±4 kg at 120 kg) and there are no weight-class jumps. Where data is thin, the window widens until at least 1,500 lifters effectively count. Lifts tied with yours count half. Tested against held-out lifters, the predicted percentiles come true to within about one point.
This ranks you among people who compete, a strong crowd. Among gym-goers in general you'd rank higher. Past the 99.99th percentile there are only a handful of people, so the page stops there.