“Ninety percent of your max” sounds arbitrary until you see what it is ninety percent of — and what happens to that number twice before it reaches the bar. On the Get Stronger goal, the Training Max is the number every week’s percentage is calculated from, and Sterka seeds it low on purpose. Not ninety percent of what you can lift. Ninety percent of an estimate that is already a cautious read of what you can lift.
A double discount, not a single haircut
The Training Max has a lineage rooted in Wendler-style periodization, where a working max is set below true max so the percentages off it stay honest week to week. Sterka keeps the ninety-percent multiplier but changes what it multiplies. The classical approach takes ninety percent of a tested one-rep max. Sterka takes ninety percent of an estimated one — and the estimate is conservative before the multiplier ever touches it.
That is the whole design: two discounts stacked, not one. The first is baked into the estimate. The second is the ninety-percent seed. Together they park the Training Max well under a true max, which is exactly where a number you build on every session should sit.
Where the number lands
Start from a true one-rep max you never actually test. Sterka reads your estimated 1RM off submaximal sets left two reps short of failure, which under-reads a true max by roughly three to seven percent — the two reps still in the tank do not show up in the math. So the estimate already sits near ninety-five percent of your true max. Seed the Training Max at ninety percent of that, and it lands around eighty-six percent of true. Ask the week-3 wave for a single at ninety-five percent of the Training Max, and the bar you actually grip sits around eighty to eighty-three percent of your true one-rep max — heavy enough to drive strength, well short of a max attempt.
The arithmetic
Take a lift the engine estimates at a 100 kg one-rep max. The Training Max seeds
at ninety percent of that — 90 kg — and floor-rounds down to the nearest loadable
plate, so it never rounds up past the intent. Week 3 of an intensity block asks
for ninety-five percent of the Training Max: ninety-five percent of 90 kg is
85.5 kg on paper, and no bar loads 85.5, so the engine rounds to 85 kg and asks
for it as a 1+ AMRAP — one prescribed rep as the floor, then as many clean reps
as you have, stopping one short of failure (RIR 1). At around eighty-two percent
of a true max that is usually several reps, not a grind for one.
That 85 kg is the number the whole chain was protecting. It is heavy — near eighty-two percent of a true max the program never made you test — but it is a weight you move for clean reps, not one you fail in the rack. The reps you deliver are the point: they set the next wave’s Training Max.
Why the old number wasn’t enough
The seed used to be eighty-five percent, and it was too conservative. Stacked on top of an already-cautious estimate, eighty-five percent compounded down to a week-3 single sitting near seventy-six to seventy-eight percent of true max — under the intensity range strength research treats as a genuine driver. You did the heavy week and still trained below the threshold that moves the number, for the several weeks the small per-session increments take to catch up. Raising the seed to ninety percent lifts that same single back into the eighty-percent-and-up strength zone without turning it into a max attempt. It is a deliberate correction, not a coincidence.
Why conservative still beats precise
An understated Training Max is not timidity — it is where the self-correction lives, and the ways of being wrong are not symmetric. Set the number slightly low and the cost is a little headroom you give back gradually: the week-3 set tests it, and depending on the reps you deliver the Training Max climbs, or holds when you meet the prescribed rep without a surplus and show you still own the load. Set it too high and the cost is a missed rep under a loaded bar — a failure the design would rather catch at one honest set than let compound into a block built on a number you cannot actually move.
So the gap between your Training Max and what you could technically lift today is not a flaw the engine is working around. It is the room the program leaves itself to be corrected by real reps instead of by an estimate.
What you do with it
Trust the number, and let the wave test it for you. You do not have to feel strong to the paper max, or second-guess a top set that reads lighter than you expected — that lighter-than-you’d-guess weight is the whole point, and it climbs on its own when you earn it. If you want to see where the discount lands for your own lifts, run the numbers in the Training Max calculator.