reckoner
Almost everybody rates a set on how much it hurt. The bar felt heavy, your heart rate was up, the last rep was ugly, so the number goes up. That is a rating of your afternoon rather than a rating of the set, and it is the reason effort ratings have a reputation for being useless.
The question is narrower than that, and much easier to answer honestly.
It is a count, not a feeling
The lifting version of the scale counts reps, and Reckoner uses its top end, from six to ten: every number on it is a count of what you left behind1. A ten is nothing left: the bar stopped, or it would have on the next one. A nine is one more rep. An eight is two. A seven is three, and a six is four, at which point you are warming up rather than working.
That is the whole of it. It is not a score out of ten, it is not a measure of how much you sweated, and there is no credit for a big number. Two people can put the same weight on the same bar for the same reps and honestly report different ratings, and they should, because the question is about them and not about the bar.
Reckoner counts good reps. On the watch the question reads Good reps left?, and on the web it asks whether you could have done more with good form. A rep you would only have got by letting the movement change is not a rep you had left.
The number is not how it felt. It is what you left.
The last set, at the moment it ended
Rate the last set of that lift, not the average of the lot and not the session. The earlier sets were done with more in the tank by design, so pooling them tells you nothing about where the lift actually finished.
And rate it standing there, before you have walked to the water fountain and decided you had another one in you. There is a measured reason to rate the end of a set rather than the middle of one: in a study of 81 men and women with a broad range of lifting experience, an estimate made within five reps of failure was out by about one rep, and one made seven to ten reps away was out by more than two2. The closer the set is to finished, the better anybody is at counting what is left3.
What Reckoner does with the number
This is the real engine, run against a lifter on the same bench press who finishes every set that is asked for, every week. What changes from row to row is the rating.
| Week | Prescribed | What it did |
|---|---|---|
| 1 | 4 x 10 at 60 kg | First week Where the run starts. Nothing is logged yet, so nothing has changed. |
| 2 | 4 x 8 at 62.5 kg | Load up 60 kg → 62.5 kgYou hit the top of the range with one rep left in the tank (RPE 9), so the bar goes up 2.5 kg to 62.5 kg. The weight went up, so the reps come down to eight. You are now working 6 to 10. You finished every set that was asked for, which is what earns it. |
| 3 | 2 x 8 at 62.5 kg | Eased 4 → 2 setsThis exercise has been over what it is asked for two trained weeks running, so Reckoner takes about 40% of its sets off for one week rather than waiting for the block to do it. The load holds and your reps stay where you earned them, so you keep the strength and lose the fatigue. Nothing else in your week moves. Rest goes up to 3:30 next week, after two hard weeks. |
| Next | 4 x 9 at 62.5 kg | More reps 8 → 9 repsYou finished every set with three reps left in the tank (RPE 7) and room left in the range, so the reps climb before the load does. The lighter week is over, so this is back to four sets. |
Real engine output, not an illustration. Every row was produced by running the engine itself when this page was generated.
In the first row the reps were all there, so the bar goes up. That is the part the rating does not decide: work that was finished is work that was done, however it was rated, and an athlete who trains at the top of the scale on purpose is not somebody the engine should freeze for a month. What the rating decides is whether you are owed relief, and it is already counting.
The middle row is what happens when a hard number keeps arriving, and two separate things land on the lift that reported it. Once a lift has run over its own ration for two trained weeks running, that lift backs off for one week rather than waiting for the block to come around to it: about 40 per cent of its sets come off, and the load and the reps stay exactly where you earned them. Underneath that, it gets 60 seconds more rest between its sets. Both land on the one lift you have been grinding, and nothing else in your week moves for it.
Your target is not one number
Every lift carries its own target and it is printed beside the lift, and above the target sits a harder line called the ration: the most this class of lift is ever asked for. Anchors and the working sets under them ration at 8, two reps in reserve, because that is where you can hold a position and repeat a week. One top set, where a day has one, rations at 9, and so does an accessory whose range tops out above twelve reps, because a long set has to be taken closer to the end to do its job. Report above your lift's ration two trained weeks running and the engine reads it as a lift asking for recovery rather than as a lifter trying harder.
Your first week holds everything to the anchors' target, accessories included, while Reckoner works out what you can lift. Nothing runs closer to the edge than that on loads that are still a guess.
Four ways the number goes wrong
Rating the day instead of the set
A bad night, a skipped lunch and a late train do not change how many reps were left in the tank. If the set was genuinely there, it was there. Answer the How are you? question if you woke up rough, because that is the place the engine wants to hear about your day.
Being brave with it
Shading a hard set down because a high number feels like an admission is the costliest mistake here, and it is the commonest. The engine believes you, so the extra rest and the lighter week you were owed never arrive, and next week is worse. An honest hard number buys you that lighter week, which is the thing you actually wanted.
Calling a grind something other than the top of the scale
If the bar slowed to a crawl and the next rep was not happening, that is the top of the scale, whatever the plan asked for. It takes nothing away from reps you actually finished, but it is counted, and two trained weeks of it on one lift earn that lift more rest and a lighter week. If the grind changed the movement, say so when you are asked about the form: fair on a set rated nine or ten, or rough at any rating, holds the reps where they are rather than asking for one more.
Rating it from memory
Logging four lifts on the train home turns four counts into one impression of the session, and an impression is not what the engine asked for. Rate each lift as you finish it, in the gym.
None of this asks you to be precise. Nobody knows whether they had one and a half reps left, and the engine is not built on the assumption that they do. It asks you to be honest about roughly how close you came, every week, in the same way, which is a much smaller ask and is the only input it cannot get from anywhere else.