Home / Guides / Strength Plateau Detection
Strength Plateau Detection: Reading the Trend, Not the Streak
Plenty of apps promise to flag your plateau. The idea is good; the usual math is not. A "your estimated 1RM has not gone up in three workouts" alert runs a streak counter on a reps-only 1RM formula, and both halves of that are wrong for anyone who trains in blocks and cycles.
The problem with "no new best in three workouts"
A plateau alert has two jobs: estimate your one-rep max from a working set, and decide when that estimate has stopped moving. The common version does the first with a reps-only formula and the second with a streak counter: if the estimate has not set a new high in three sessions, it fires. Both choices break down the moment your training becomes structured.
Reps-only estimates are blind to effort
The reps-only estimate almost everyone uses is the Epley formula:
Notice what is missing: how hard the set was. Grind out 235 for 9 at RPE 9, or bang out 235 for 9 with three reps left in the tank at RPE 7, and Epley returns 306 either way. So imagine a lifter who hits 235 × 9 every week for two months while the set gets steadily easier, RPE falling from 9 down to 7. That is real progress, the same weight moving with more in reserve, and a reps-only estimate reads a flat 306 the entire time and calls it a plateau. The reverse is just as bad: if the same weight and reps start costing more, RPE climbing week over week, that is a regression a reps-only number renders as "no change."
An RPE-aware estimate fixes this by using proximity to failure. The same 235 × 9 implies a much higher max at RPE 7 (reps in reserve) than at RPE 10 (nothing left), so the estimate rises as the lift gets easier and falls when it gets harder, which is exactly what a plateau read should track.
A streak fires during normal training
The streak counter is the second flaw. Undulating and percentage-based programs are built so you do not set a rep max every week. Back-off sets, volume blocks, deloads, and the down weeks of a wave all produce runs of "no new best" by design. A counter that flags those runs is flagging the plan working as intended, which is how you train a lifter to ignore the alert. Strength genuinely plateaus as a normal part of periodization, so a flat week is information, not an emergency.
How to detect a real plateau
A plateau is a property of the trend, not of the last few sessions. To read it honestly:
- Estimate with effort in the number. Use an RPE-aware estimated 1RM so a lift getting easier reads as progress and a grind reads as a stall.
- Fit a robust trend, not a streak. Take your best estimated 1RM per session over roughly the last eight weeks and fit a line through it. A robust fit (the Theil-Sen estimator, the median of every pairwise slope) means one sick-day session or a data-entry typo cannot swing the verdict the way it can under ordinary least squares. Flat or falling over the window is a plateau; clearly rising is not.
- Require enough data. A trend needs enough recent sessions spread over enough calendar time to mean anything, and a lift you have not trained in weeks is being detrained, not plateaued. Below that floor, the honest answer is "not enough data," not a false alarm.
Grind, or just volume?
There is one more signal that turns a flat estimate into a diagnosis: your RPE trend on the same lift. A flat estimated 1RM while your RPE is climbing is a genuine grind-stall, harder work for the same output, and it is the least ambiguous plateau there is. A flat estimated 1RM at an easy, steady RPE is usually just light or volume work that was never meant to set a max. Reading effort alongside the estimate is what separates the two, and it is invisible to any tool that only counts reps.
Reading your own trend
- If your estimated 1RM is flat but the weight feels easier (RPE drifting down), you are getting stronger. Keep going; the number will catch up.
- If your estimated 1RM is flat and RPE is drifting up on the same loads, that is a real stall. A back-off or a planned deload usually gets it moving again.
- If it is flat because you are mid-accumulation or on a down week, that is the plan. Judge the plateau at the block level, not the session level.
- If a lift only went flat because you stopped training it, that is detraining, not a plateau.
Ægir Iron does this automatically. It tracks the trend of your RPE-aware estimated 1RM per lift over the last eight weeks with a robust Theil-Sen fit, reads your RPE trend alongside it to tell a grind from easy volume, and surfaces a flat or falling lift in your Analytics summary as a description of the trend, never as an order to push. The full method, with citations, is documented in the app.
Common questions about strength plateaus
How do I know if I have really plateaued? A real plateau is a flat or falling trend in your estimated 1RM over a meaningful window, not a few workouts without a new best. Because programs wave loads on purpose, short runs of no new best are normal. Fit a robust trend over about eight weeks and read your RPE alongside it.
Why is my estimated 1RM flat when the weight feels easier? A reps-only formula like Epley reads the same weight and reps as the same estimate every time, no matter how hard the set was. An RPE-aware estimate uses proximity to failure, so it rises as the same lift gets easier.
Is a plateau alert after three flat workouts reliable? Not on its own. Three workouts is too short to separate a plateau from a normal training wave, a deload, or a volume block, all of which produce runs without a new best by design.
What app detects strength plateaus? Ægir Iron flags a flat or falling trend in your RPE-aware estimated 1RM per lift over the last eight weeks with a robust Theil-Sen fit, and reads your RPE trend alongside so a grind is not confused with a plateau.