PG-ART
Performance-Gated Autoregulated Resistance Training: a set of rules that reads your logged reps and decides, on its own, whether your next target goes up, holds, or comes down.
Y-axis is training volume (weight × reps). At a fixed weight, volume climbs as reps go 6 → 10; clear the top and the weight steps up while reps reset to 6, so volume briefly drops before each cycle climbs from a higher baseline.
What it is
PG-ART is an algorithmic workout progression: you train four sets per exercise in a fixed 6-10 rep range, and a small set of rules turns the reps you logged into your next session's target. There is no RPE or RIR to estimate; the progression is decided from objective rep data, so the same log always produces the same next step.
The problem it solves
Most progression schemes fail in one of two ways. Fixed calendar programs (add weight every week regardless of how the sets actually went) eventually schedule a jump you haven't earned, and you grind, miss, or ego-lift through it. Subjective autoregulated resistance training (adjusting load by RPE / reps-in-reserve) fixes the rigidity but replaces it with guessing: "how many reps did I have left?" is exactly the judgment call that drifts on a bad day and stalls you on a good one.
PG-ART keeps the responsiveness of autoregulation but removes the guess. Performance, meaning did you hit the reps or not, is the only input, and a gate decides the rest.
The rules, stated precisely
The whole method is five rules operating on one exercise at a time.
- Format
- Rep range 6-10. Four sets per exercise, about 90 seconds of rest between them. A four-exercise session runs roughly 45 minutes.
- Success gate
- Hit the target reps in 3 of the 4 sets and the gate passes, so next session the target for that exercise rises by 1 rep.
- Step-down
- Fail the gate (target hit in fewer than 3 sets) and the target drops by 1 rep, the smallest proportionate correction, so you re-approach the same number instead of restarting.
- The Volume Sawtooth
- Pass the gate at the top of the range (target 10 reached) and the weight goes up about 5-10% while the target resets to 6. Intensity steps up; total tonnage drops for a session or two before climbing again. That reset is a built-in functional deload.
- Floor self-correction
- Fail at the bottom of the range (target 6) and the weight itself is too heavy, so the load is reduced. The floor catches a jump that turned out to be too big.
Because every branch is defined, the method is fully algorithmic: log a set, the rules set your next target. Nothing depends on how the set felt.
Where it sits among existing methods
PG-ART is not invented from scratch. Its rep-then-load ladder is double progression: add reps until you top the range, then add weight and start the reps over. Its gate is a performance test in the spirit of APRE-style autoregulation, where the set result, not a calendar, chooses the next load. What PG-ART actually contributes is a specific, fully automated combination of the two: a fixed range, a 3-of-4 gate, a single-rep correction, and a defined reset, tight enough that a double-progression app can run it end to end with no human judgment in the loop.
My 10 months of data
I ran PG-ART on myself and logged every set. This is a self-experiment, a feasibility record of what the rules did for one lifter over time, not a study, not clinical evidence, and not proof it will do the same for you.
Aug 2025 - Jul 2026
Working-weight change over that window:
- Chest press +18.2% (55 → 65 kg)
- Leg extension +18.2% (55 → 65 kg)
- EZ-bar curl +32.6% (23 → 30.5 kg)
- Leg curl +9.1% (55 → 60 kg)
Not everything moved. One lift, rows, plateaued after I changed the exercise order mid-run, and it never fully recovered its progression in this window. I'm keeping it in the record rather than hiding it: it's a clean example of the method's own precondition (a stable exercise order) being violated, and the measurement showing it.
Two numbers say something about how the gate behaves in practice. In 36% of complete sessions the pass/fail was decided by the fourth "buffer" set, so the gate earned its 3-of-4 design, and 28% were clean sweeps where all four sets hit.
Where it's meant to be used
These conditions are part of the method, not fine print. PG-ART assumes machine or fixed-path exercises (where a rep is a rep and load is easy to nudge by small amounts), a stable exercise order so fatigue lands the same way each session, and stable rest so the reps you log are comparable across weeks. Change those and the gate is reading a moving target, as the rows plateau above shows.
Iron Log runs PG-ART for you
You don't work the rules by hand. Iron Log is the local-first Android tracker that applies the gate, the step, and the sawtooth reset automatically: you log the set, it sets your next target.