How does Base decide the weight for my next set?

From your last session on that exercise, applying double progression with an RIR clause. Every set at the top of the rep range at RIR 1 or below earns a load increase and a reset to the bottom of the range; inside the range means hold the load and add a rep.

What is an e1RM trend and why does it matter?

An estimated one-rep max converts sets at different weights and rep counts onto a single axis, so a block's worth of messy training data becomes one line. The slope of that line is the honest answer to whether the block is working, and it is one of the four signals that trigger a deload.

Does Base calculate warm-up sets?

Yes, ramped from the bar to your working weight in a sensible number of steps, with plate maths for each. A warm-up that has to be worked out in your head is a warm-up that gets skipped or overdone.

How accurate is an estimated 1RM?

Within a few percent for sets of one to five reps, degrading above that and unreliable past about twelve — which is why Base refuses to compute one above twelve reps. High-rep sets over-predict maximal strength because they are limited by local endurance rather than peak force.

Can I see how an exercise has progressed over months?

Yes. The exercise detail view holds the full history for that movement — every set, its RIR, and the e1RM trend across the whole period — which is the view that answers whether a plateau is real or is one bad session in a rising line.

The Exercise Logger: Everything About One Movement

The e1RM trend

Across a training block your working sets vary for reasons that have nothing to do with strength: a deload week, a bad night's sleep, a change in exercise order, a session cut short. Comparing individual sessions to each other is comparing noise.

An estimated one-rep max converts all of them onto one axis:

e1RM = weight * (1 + reps / 30)

Epley, and only for sets of twelve reps or fewer. Above that every 1RM formula degrades and it degrades in a direction — high-rep sets over-predict maximal strength, because a set of twenty is limited by local muscular endurance and buffering capacity rather than by peak force production. Base refuses to compute above twelve rather than showing a number it does not believe.

RIR goes into the input. A set of five at 225 with two reps left is a different data point from a set of five at 225 taken to failure, and a formula fed only "225 for 5" cannot tell them apart. Adding RIR to the rep count before estimating is what makes sets at different efforts comparable.

The slope is the information, not the number. Nobody should quote an e1RM as a max. What it is genuinely good for is the line: 0.5 to 2% per week early in a block, tapering. Flat for three weeks alongside a second fatigue signal is a deload trigger, not a programming failure. The calculator.

The warm-up ramp

Calculated from the bar to your working weight in a sensible number of steps, with plate maths for each.

This is a small feature that removes a real friction. A warm-up you have to work out in your head is a warm-up that gets skipped when you are in a hurry or overdone when you are not paying attention, and neither is free — an insufficient warm-up on a heavy compound is where the RIR you were aiming for stops being achievable.

Plate maths matters more than it sounds for the same reason. Standing in front of a bar doing arithmetic is a small tax on every set, and the app already knows the target.

Per-exercise history

Every set on this movement, with its RIR, in order.

Two things this answers that a session-level view cannot:

Is this plateau real? One bad session in a rising line is a bad session. Three sessions at the same load with RIR falling is a stall. Those look identical from inside a single workout.

When did the load last move? If a lift has held the same weight for five weeks, the progression rule has been saying "hold" for five weeks, and the reason is either that you have not been hitting the top of the range or that your RIR has been above 4. Both are diagnosable from the history and neither is visible from today.

Aliases, and why they exist

The same movement gets called different things — "barbell row", "bent-over row", "BB row" — and a log that treats those as three exercises has three histories with a third of the data each, and no usable trend on any of them.

The exercise catalogue resolves aliases so that history follows the movement rather than the spelling. Variations that are genuinely different lifts are kept separate: a Romanian deadlift and a conventional deadlift are not the same exercise, however similar the name.

Fractional set volume

Each set on this exercise contributes to the weekly volume count for the muscles it works — in full for the muscle acting as prime mover, and at half for the muscles assisting.

That fractional model is the one that best fit the hypertrophy dose-response data in Pelland et al.'s meta-regression across 67 studies and 2,058 subjects. A barbell row is one direct set for the lats and half a set for the biceps; a close-grip bench is direct for the triceps and half for the chest.

It matters for a hybrid athlete because the productive range shifts with endurance load — 14 to 20 fractional sets per muscle per week with no endurance work, 12 to 15 at four to six hours a week — and counting only direct sets systematically under-reports where you actually are. The calculator.

What the screen refuses to do

No estimated 1RM above twelve reps. The formula does not support it.

No percentile against other users. A comparison of your squat to other people's ranks their bodyweight, their training age and their leverages. Your own trend is the reference that means something.

No form score from video. Consumer pose estimation is not good enough to judge a bar path, and a confident wrong answer about your technique is worse than no answer.

What to do next

The workout screen is where sets get logged, and the program editor is where the rep ranges and RIR targets these rules run against get set.