The Endurance Screen: Prescribe the Session, Then Grade It

Nothing works before there is a threshold

Until a threshold exists for at least one discipline, the screen shows one thing: a prompt to set one. No zones, no charts, no partially useful placeholder.

That is not minimalism. Every band on this screen is a percentage of that number, so a screen full of bands built on nothing would be a screen full of confident, wrong prescriptions. Base can estimate the threshold from what you have already run and ridden, or you can enter a recent test result — and a manual override always wins, because a real race or a proper field test is better than anything inferred from training data.

How threshold is estimated

Threshold is roughly the hardest effort you could hold for about an hour. It is the dividing line between work you can sustain and work that accumulates fatigue faster than you clear it.

For running, Base takes your best sustained effort in the window and projects it to a one-hour equivalent using Riegel's formula. That projection is not optional: a twenty-minute effort is genuinely faster than hour pace, and using its raw pace would set every zone too fast — which lands hardest on the easy days, the exact place where being too fast does the most damage.

For cycling, it is the classic field test generalised: average power scaled by how long the effort already lasted, reaching a factor of 1.0 past about fifty minutes. A flat 95% discount is right for a twenty-minute test and wrong for a forty-five-minute climb.

Both need at least twenty minutes of sustained work. Below that the projection says more about the formula than about you.

Each estimate carries a confidence label — high, medium, low — based on how recent and how long the effort was. Low means the estimate is old or the effort was short, and every band underneath it should be read accordingly.

Run and ride never mix. They are different units measuring different capacities, and a combined "endurance score" could not answer either question.

The bands

Run paces, as multiples of threshold pace. A larger multiple is a slower pace:

Ride power, as fractions of FTP — the standard Coggan bands, because that is what every cyclist and every head unit already means by "zone 3":

Easy and long runs also carry a heart-rate band at 60–70% of your age-estimated maximum. On an easy day, obey the heart-rate band; on interval days, obey pace or power. Pace drifts several percent with heat and hills while heart rate does not — and on a three-minute rep heart rate lags the effort by so long it is useless.

If no birth year is on file there is no honest maximum to take a percentage of, so no zones are shown at all rather than an age being assumed.

Today's Target

The session you chose, in numbers you can set on a watch, with the interval structure if it has one and the watch setup line so you can transfer it in a single read.

The band is frozen onto the row when the session is created. Re-estimating your threshold tomorrow does not rewrite what today asked for. If it did, every past grade would silently become a grade against a session that was never prescribed, and the training history would stop meaning anything.

Tap "heading out" before you leave. That timestamp is what lets Base match the right workout back to this prescription when you run twice in one day. Without it, two runs and two targets are a coin toss — and Base will refuse to guess rather than attach the wrong grade to the wrong session.

The grade

This is the feedback loop that makes prescribed training worth anything, and it is the single most valuable thing on the screen for a hybrid athlete.

Three levels of evidence, best first, and the card always states which one it used.

Per-rep. A structured session is graded rep by rep against the laps your device actually recorded — never against intervals re-derived from a pace stream, because the watch already marked them exactly and re-deriving them introduces error where there was none.

Time in zone. A continuous session is graded on how much of its measured time fell inside the band. This is the one that matters most, and the reason is specific: an average hides the run that went out too hard and limped home. The two halves cancel, the mean lands comfortably inside a band the run was almost never in, and the session grades as a success. Time-in-zone catches it. An average structurally cannot.

Average. Only when neither of the above is available, and it says so.

"Nailed" is 70% of reps or 80% of time in band. A few percent outside counts as "close", because that is inside normal GPS and power-meter error and grading it as a miss would be pretending to a precision the data does not have.

The wording is deliberately not punitive. "Under" on an easy day is usually correct behaviour, not a failure. What matters is the pattern. Repeatedly "over" on easy days means the easy days are not easy — the most common reason a training block stops working, and completely invisible without this comparison. Repeatedly "under" on intervals means the prescription is too aggressive for current fitness.

Ramp rate

This week's endurance minutes divided by the average of the previous four. It sits at the top of the screen, above everything else, because it is the one number here worth acting on immediately.

The 0.8 to 1.3 window is a productive build. Past about 1.5 the injury-risk curve turns sharply upward. Below 0.8 is reported as neutral rather than as a warning — a taper week and a rest week both live there, and both are deliberate.

Total minutes is a deliberately crude measure of training stress. It needs no power meter, no heart-rate strap and no lab test, and the thing it is catching is a large sudden jump, which is coarse enough that a coarse measure finds it. A better metric with holes in it is worse than a crude one that is complete.

For a hybrid athlete the ratio does double duty. A spike week is the classic setup for an overuse injury, and it also draws directly on the recovery budget your lifting is spending. The interference between the two is much smaller than folklore claims, but it is not zero when endurance volume jumps.

What this screen does not do

It does not run the session. Apple Watch and Garmin both handle live pacing, zone alerts and structured intervals natively and well, and building a worse version of that would help nobody. Base stops at deciding what the session should be, and starts again at telling you honestly whether it landed.

What to do next

Setting your threshold covers the estimate, the confidence label and when to override it by hand. Why your easy runs are too fast covers the failure this screen exists to catch.