Why your e1RM calculator does not know your athlete

A rite of passage for the budding lifter is constantly checking an estimated one rep max (e1RM) calculator after every top set. But have you ever stopped to think about what is actually behind those numbers: who built the formula, on whom, and why?

The Epley and Brzycki charts are two of the most commonly used tools people reach for to guess a top end single. These simple, decades old formulae are dinosaurs that were never built to reflect any one athlete's individual traits. Epley's formula is e1RM = weight × (1 + reps / 30). Brzycki's is e1RM = weight × 36 / (37 − reps).

As you can see, both are simple, and both only really hold up well at low rep ranges and for an "average" lifter. Neither has any way to tell a heavyweight from a lightweight, and both are completely blind to how you, specifically, scale as you approach your max. That is by design: they are meant to be a rough and ready population guesstimate, one that averages out every individual quirk and abnormality into a single distilled number. The modern research suggests that is not the best way to do this.

Knutzen et al. (1999) tested six e1RM equations against real 1RM results (in older adults, on machine exercises, so not a like for like match to a competitive lifter on a barbell, but the pattern still holds). They found a "moderate to strong relationship for all exercises (upper extremity: r = 0.77–0.90; lower extremity: r = 0.60–0.80)". That sounds reassuring until you square it: an r of 0.90 only accounts for about 81% of the variance in the real result, and the lower extremity numbers, at r = 0.60, explain barely a third of it. Six different equations, and none of them close the gap.

Richens and Cleather (2014) found that an athlete's training background changes that gap further. Comparing eight competitive weightlifters against eight distance runners on the leg press, the runners completed more than double the reps of the weightlifters at 70% of 1RM (39.9 vs 17.9) and around 70% more at 80% (19.8 vs 11.8). That difference disappeared by 90%, right where the two groups were no longer significantly different. In other words, the two populations diverge most in exactly the mid rep ranges most e1RM tests actually use, and converge only once you are already close to a true single.

A recent, not yet peer reviewed preprint by Marzagao (2026) makes a related but separate point. Built from over 300,000 real training sets across nearly 400 exercises, it found that letting the reps to 1RM conversion vary with how much weight is actually on the bar, rather than assuming the same fixed relationship no matter the load, cut prediction error by 17 to 22% against Epley, Brzycki, Wathen, and Mayhew. An ablation attributed 91% of that improvement to the weight dependent term and the remaining 9% to a smaller tweak in how reps themselves are weighted. This is not itself a personalised model: every lifter doing that exercise still gets the same formula. But it shows that even a one size fits all approach is measurably wrong once you account for the load and the exercise, which is exactly the kind of context an individualised fit, built from one athlete's own bar, can pick up automatically.

You would not program a 43kg lifter the same way as a 120kg+ lifter. You would not compare their top sets and working sets without understanding the differences between them. And you would certainly not put together an attempt sheet without real care for the fact that you have two very different athletes in front of you. So why not individualise your one rep max calculator too?

That is exactly what Auriga does. Our training analysis tools adapt and learn from your athletes' own lifting, to give bespoke, individualised insight into their strength and their development. From measuring strength to estimating loads, and from tracking e1RMs to planning an attempt sheet, the guidance Auriga offers is tailored to the athlete in front of you, not an average of everyone who has ever touched a bar.