How the difficulty bands are calculated
Beginner, Advanced and Elite come from a handful of pace rates and a per-exercise multiplier. Here are the actual numbers, the arithmetic in both directions, and what the model still does not account for.
While you build a workout, each station carries a target time for three levels: Beginner, Advanced and Elite.
There is one idea underneath all of it.
Every station reduces to an amount of work and a rate, and a band is just a rate.
Multiply the work by the rate and you have a time; the three bands give you three times.
Work comes in three currencies, so there are rates for each.
Repetitions are priced in seconds per rep and calories in seconds per calorie.
Distance is priced three ways, because a metre is not one thing: running, ergometers, and everything else (sled work, carries, anything you move rather than cover).
The decimals are easier to judge in familiar units.
On repetitions the bands work out at 20, 25 and 30 a minute, and on calories at roughly 10, 14 and 16 a minute.
Running comes out at 7:00, 5:20 and 4:00 per kilometre.
Rowing, priced separately, lands at 2:35, 2:22 and 2:11 per five hundred metres.
The bands sit closer together there than anywhere else on purpose: an ergometer is machine-paced, and across a real race field the gap from the median athlete to the top one percent on the ski and the rower is about fifteen percent, against seventy on a sandbag lunge.
Spreading the erg as wide as everything else had put our Elite a quarter faster than the fastest people in the sport.
If those look like the right shape for three levels, the model is doing its job.
Not every repetition costs the same, so a flat seconds-per-rep would be wrong in an obvious way: a burpee is not a wall ball.
Each exercise belongs to one of four classes, and the class carries a multiplier: fast at 0.6, moderate at 1.0, slow at 1.6, and grind at 2.2.
Air squats, sit-ups and kettlebell swings are fast.
Thrusters, box jumps and pull-ups are moderate.
Burpees and sandbag work are slow.
Devil’s press, man makers, wall walks and the sled are a grind.
Those four numbers are not preferences.
Each class is anchored to an observed working cadence, with the moderate class set at the baseline rep rate of a thruster or a box jump, roughly 2.4 seconds.
A wall ball or air squat runs nearer 1.4 seconds, which is where 0.6 comes from.
A burpee around 3.8, giving 1.6.
A devil’s press around 5.3, giving 2.2.
The cadences come from published HYROX and CrossFit movement data rather than from how hard something feels.
Bar width is the multiplier; the right column is what twenty repetitions cost at Advanced pace. Same count on the whiteboard, nearly four times the clock.
The general distance rate was set the same way and was the one that had been most wrong.
It had been priced like running, which badly under-costs moving weight: sandbag lunges run somewhere between 1.5 and 3.0 seconds per metre, so a hundred metres is two and a half minutes for a strong athlete and five for a recreational one.
That is now its own rate rather than a copy of the running one.
There is a check on all of this that does not depend on anyone’s judgement.
Every completed session already records how long each station took and how many repetitions it held, so the observed cadence per exercise is a measurement, not an estimate.
A tool in the repository aggregates it (the median for an exercise against the median across all of them) and proposes the multiplier the data implies.
It uses medians rather than averages so one sandbagged station cannot drag a number, requires a minimum number of observations before trusting an exercise, ignores stations logged as distance rather than reps, and refuses to write anything itself.
It prints a proposal for a person to accept or reject, because data drawn from your own users carries its own biases and the decision to publish should be visible in the history.
Two movements already carry a refined rate on this site, on top of their class, because the evidence demanded it.
Wall balls price at 1.2x rather than the fast class’s 0.6: this season’s race results put top-ten-percent athletes near 2.95 seconds per ball on station eight, and no fresh-gym discount stretches that to the 1.4 seconds the class implied.
A hundred wall balls now price at 6:00, 4:48 and 4:00 across the bands.
Jump rope moves the other way, to 0.25, because singles genuinely run 80 to 120 per minute and no counted class is that fast.
A third refinement priced farmers carry as its own thing, at about a third of the general distance rate, because a loaded jog is nothing like the grinding sandbag lunges the general rate was built on.
The app shows the class targets until it adopts the same table.
The calculator and the exercise pages on this site already use the refined rates.
The difference is not subtle.
Twenty air squats come out at 0:36, 0:29 and 0:24 across the three bands.
Twenty burpees, the same count of the same currency, are 1:36, 1:17 and 1:04.
Twenty devil’s press are 2:12, 1:46 and 1:28, nearly four times the air-squat figure for an identical number on the page.
A timed station runs the arithmetic backwards, and this is the more useful direction.
You have already fixed the window, so the question is how much work fits inside it.
Sixty seconds of rowing is 200 metres at Beginner, 250 at Advanced and 316 at Elite.
That is why a timed station shows a distance, a rep count or a calorie count rather than a time: the time is decided, and the target is what you are asked to produce within it.
The workout style changes the question again.
An AMRAP has a cap, so a band is expressed as rounds completed inside it.
An EMOM asks something simpler and more brutal: at this rate, does the work fit in the minute at all?
A station that takes fifty seconds at Elite and seventy at Beginner is an EMOM for one of them and a treadmill of accumulating debt for the other.
All of these numbers are served from a file rather than compiled into the app, so anything that turns out to be wrong can be corrected for everyone without an App Store release.
The app ships with the same values built in and falls back to them if the file is missing, malformed, or has a band out of order, so a bad edit means the estimates stop changing rather than start lying.
Two limits worth stating.
A class is a coarse instrument: every exercise inside it is priced identically, so the model separates a burpee from a wall ball but not a burpee from a sandbag clean.
And the multiplier is assumed to hold across all three bands, meaning an Elite athlete is faster at burpees and wall balls by the same proportion.
That is an approximation, just a much smaller one than pretending the two cost the same.
You can run all of this arithmetic yourself: the band calculator prices any workout you sketch, or the full race, against the same published rates. Free, no login.
In the app, these targets appear live while you build a workout, as part of Rep The Set Pro.
One thing the numbers are not.
A band is a pace, not a verdict.
Finishing at the Advanced line means you moved at that rate on that day, which is useful to know and a poor thing to take personally.