Worked example: 315 lb for 5 reps
The seven standard equations return 354, 354, 358, 367, 368, 370 and 375 lb. Median 367 lb, spread 21 lb, about 6%. That is roughly the difference between two plates and a 25 a side and adding a further 10 — real, but not the difference between a good training block and a bad one.
What is different about squat rep sets
Squat sets sit between bench and deadlift in how quickly they degrade. Breathing and bracing become the limiting factor before the legs do on high-rep sets, so a set of ten squats tends to under-represent a strong single by more than the same set of bench would. At two to five reps the estimate behaves much like it does on any other lift.
Depth is the other variable no formula can see. If your fives are to parallel and your single would be a quarter squat, the estimate is meaningless — and if it is the other way round, the estimate is low. Keep the depth constant and the number tracks your progress honestly.
Depth is the variable, not the formula
Two people can enter the same weight and the same rep count into this page and be describing lifts that differ by 30%. The formula cannot see depth, and depth is the single largest source of disagreement between one squat number and another.
The convention worth holding yourself to is the competition standard: the crease of the hip below the top of the knee. It is unambiguous, it is what every published standard assumes, and it is the depth at which your squat and somebody else's are the same lift. A squat to parallel is a different exercise with the same name, and a quarter squat is a different exercise again.
The other invisible variable is bracing. A squat set usually ends when the trunk can no longer hold position rather than when the legs give out, which is why squat rep sets sit between bench and deadlift in how well they predict a single: better than a deadlift, where grip and lower back arrive first, and worse than a bench press, where the failure is cleanly in the prime movers.
Methodology
Seven published one-rep-max regressions — Epley (1985), Brzycki (1993), Lombardi (1989), O'Conner (1989), Wathan (1994), Lander (1985) and Mayhew (1992) — all applied to your set, with the median reported. They were fitted mostly on bench-press data in college athletic populations, which is the main reason to treat a squat estimate as slightly looser than a bench one. Full detail on how these calculators work.
Limitations, stated plainly
These formulas assume a set taken close to failure with consistent technique, in the two-to-ten rep band. Above ten reps every one of them drifts and they disagree about the direction. Treat the output as a programming number, not a target to attempt cold: warm up thoroughly, work up in small jumps, and set the safety bars at a height that lets you fail without being pinned.
Nothing here adjusts for bar position, stance, footwear or whether you wore a belt — and all four move the number. A high-bar squat in flat shoes without a belt and a low-bar squat in heeled shoes with one are not the same lift, and the gap between them for the same person is frequently larger than the gap between the highest and lowest of the seven formulas. Compare your estimate against your own previous estimates, taken the same way, rather than against anybody else's.
Treat this as a starting point for programming, never as a number to attempt cold. Squat in a rack with the safety pins set at a height that lets you fail without being trapped, or with a spotter who knows how to take the bar. This is a fitness estimate, not medical advice, and not a substitute for a professional who can actually see you.