How the Strength Score is made.

The formula does not read the bar. It reads what the bar implies: the most you could lift once, estimated from the best working set of the session by the Epley formula, weight × (1 + reps ÷ 30), with a single rep taken as read. Only completed working sets count. A warm-up is a warm-up.

Then it is smoothed. The newest session counts for forty percent of the estimate and everything before it for sixty, so one heavy day cannot whipsaw the number and one flat day cannot sink it. Per slot the strongest movement you have is the one that is read — a barbell bench and a dumbbell bench are the same slot, and the higher estimate stands for it.

A bodyweight movement is a load like any other. A rep moves a share of you through the working joints, and the app knows the share: a pull-up moves all of you, a dip about ninety-five percent, a push-up about two thirds. Bodyweight × share, plus anything you added, is the weight; the reps are the reps; Epley does the rest. This is what puts a calisthenics athlete and a barbell lifter on one axis.

A plain ratio of load to bodyweight quietly favours lighter people. Muscle strength grows with the cross-section of the muscle, and cross-section grows with the square of a body’s dimensions while mass grows with the cube — so strength rises with roughly the two-thirds power of bodyweight. Dividing by bodyweight punishes a 100 kg lifter twice. Dividing by bodyweight to the 0.67 puts a 60 kg lifter and a 100 kg lifter on one axis, which is the allometric case Slobodan Jaric set out in Sports Medicine in 2002.

The app writes it as load ÷ bodyweight × (bodyweight ÷ 80)^0.33, which is the same thing anchored at 80 kg so the numbers stay readable. One anchor for everyone — it is not split by sex, and it is not a target. The pair under this paragraph is the term doing its job: two squats the plain ratio calls eighteen percent apart, level.

A deadlift is naturally a high multiple of bodyweight and an overhead press a low one. Averaged raw, the deadlift would drown the press and the score would mostly be a deadlift score. So each movement’s relative strength is divided by that movement’s own “advanced” standard — a multiple of bodyweight, per lift, per sex — and reads as a fraction of it. Strong-for-press and strong-for-deadlift both come out near 1.0 and count the same.

Sex enters the formula here and nowhere else. Where it is not given, the standard is the average of the two columns, never the higher bar, because dividing an unknown by the higher bar would deflate every unknown score. The worked tables by bodyweight are on the strength standards page.

Each movement’s read is multiplied by a recency factor: 1.0 the day you do it, decaying with a time constant of twenty-one days toward a floor of 0.7. Three weeks untouched reads at about 0.81; three months, at about 0.70; a year, still 0.70. The floor is the point. Strength you built does not evaporate because you stopped logging it, and the score should not pretend it did.

It is applied once, inside the formula, per movement. Nothing downstream decays anything again.

Summing would pay for volume: log a fourth and a fifth movement, however weak, and the total goes up. Averaging pays only for quality, and it means the score of a lifter who trains two movements well is not a fraction of the score of one who trains five. The score is the mean of the normalised, recency-weighted reads of every movement you have — one to five of them.

Breadth is still rewarded. It is rewarded by the next term, gently, and not here.

The coverage factor is 0.8 plus 0.2 × the share of the five slots you have logged: 0.84 for one movement, 0.88 for two, 0.92 for three, 0.96 for four, 1.0 for five. That is the whole reward for breadth — a fifth of the score at most, in even steps, and never enough to let a weak fifth movement beat a strong fourth.

The five slots are movement patterns, not exercises. A pistol squat fills the squat slot; a hip thrust the deadlift slot; a dip the press slot; a pull-up the row slot. Somebody with rings and a floor can fill four of the five without a barbell.

Peak strength is a young adult’s. Past twenty-eight the formula applies one multiplier — 1 + 0.007 × the years past the peak — so a forty-five-year-old’s lifts are read about twelve percent up and a sixty-year-old’s about twenty-two, with a hard ceiling of 1.5 that no age reaches in practice. One bounded factor, applied once to the whole score, and nothing else in the formula knows how old you are.

Everything above it is a ratio: 1.0 is advanced at every movement you do, on your bodyweight, at your age, this week. The Liftor Number is what the ratio is multiplied by to become the number on the meter — a calibration, per lift and per lifter, that is the app’s and is sealed. It has its own section below, including what happens when you try to solve for it.