How your percentile is calculated

Every lift is compared only with the same exercise, the same gender, and the same weight class (a bucket). Your weight class is set automatically from the bodyweight you enter.

When a bucket is new and nearly empty, we lean on a published strength chart to know what "normal" looks like. As real people log lifts in that bucket, their numbers get more and more say. After 30 real entries, the live data and the chart count equally. Beyond that, the live data counts for more than the chart.

For each lift we store a chart with bodyweight rows at 155, 175, 200, 220, 240, 265 and 310+ lb. Each row lists five tiers, and each tier stands for a fixed share of lifters:

Your exact bodyweight almost never lands on a row. So we take the two nearest rows and blend them in a straight line. At 187.5 lb you get the halfway point between the 175 and 200 rows. Below 155 lb or above 310 lb, the end row is used.

Those five numbers are then turned into a bell curve. We draw the straight line that best fits "tier value vs. z-score" (least squares), and that line gives us the chart's average (μ_standard) and spread (σ_standard):

value ≈ μ_standard + σ_standard · z

μ_standard = (B + N + I + A + E) / 5
σ_standard = Σ zᵢ·valueᵢ / Σ zᵢ²

We also take every verified (non-flagged) entry in your bucket and work out its real average (μ_actual) and standard deviation (σ_actual). Let n be how many entries that is. A spread needs at least two lifts, so with fewer than two entries the live σ is left out.

The two are combined as a weighted average. The chart counts as if it were n₀ = 30 lifters, and the real data counts as n lifters:

μ_blended = (n₀ · μ_standard + n · μ_actual) / (n₀ + n)
σ_blended = (n₀ · σ_standard + n · σ_actual) / (n₀ + n)

n₀ = 30

Your lift is placed on the blended bell curve. Φ is the standard normal cumulative distribution function. It gives the share of the curve that falls below your z-score:

z = (weight_lifted − μ_blended) / σ_blended

percentile = Φ(z) × 100

A percentile of 90 means your lift beats about 90% of the bucket, so you're in the top 10%. Your tier is the highest tier mark that percentile clears.

A new entry is automatically flagged for review if either of these is true:

• its z-score against the bucket's blended curve is above 3 (about a 1-in-740 lift), or
• the weight lifted is more than 5× bodyweight.

Flagged lifts are left out of every percentile calculation and every leaderboard until an admin reviews them. You can still see your own flagged lifts on My Results, marked "under review".