Why we call Trailwise femtech

There is a week most months where nothing adds up. Your resting heart rate is two beats high. Your HRV is down. You sleep warm. The watch reads all of that and tells you to go easy, and hints that you might be coming down with something. You are not coming down with anything. You are eight days out from your period, and your body is doing what it does at that point in every cycle.

Then it happens again the next month. And the month after that.

Most training apps never see the pattern. The one piece of information that would explain it is the piece they never ask for.

We call Trailwise femtech because we went and asked.

The numbers were measured on someone else

Start with your maximum heart rate, because almost every zone you train in hangs off it.

If you have never done a max effort with a chest strap on, an app has to estimate. The usual estimate is Tanaka, 208 minus 0.7 times your age, derived on a mixed group that was mostly men. Trailwise uses Gulati for women, 206 minus 0.88 times your age, derived on a cohort of around 5,400 women. At 40 those two formulas are nine beats apart. Tanaka says 180. Gulati says 171.

Nine beats is not a rounding error. It is the width of a zone. Train your easy runs against the higher number and your easy runs are not easy, all season, and you will read the resulting fatigue as a character flaw.

A real observed maximum always wins over either formula. The formula is only there for the day you have no measurement, and on that day it should at least be a formula built on a body like yours.

Most of the sports science a running app ships was measured on men and handed to everyone.

That is the whole argument, and the maximum heart rate is only the cleanest example of it. Readiness scores treat a warm body and a suppressed HRV as a warning sign, whatever is causing them. Load models assume the athlete’s physiology holds steady from one week to the next. An app that never asks about a period cannot notice when one stops arriving, which is one of the most useful things a coach can notice at all.

Reading a warm week correctly

Here is what the cycle support actually does, in order.

You tick the first day of your period in the daily check-in. That is one tap, and it is the only thing the model needs to start. From a history of those onsets Trailwise estimates where you are in your cycle. The luteal phase, the stretch between ovulation and your next period, runs close to 14 days for most people and varies little. Cycle length varies mostly in the first half. So the estimate counts backwards from your predicted next onset rather than forwards from your last one, which is the more stable end to measure from.

Eight days before that predicted onset, a flag goes up internally called the luteal confound. In that window a temperature rise and an HRV dip are expected physiology.

Then the scores change how they read you, by specific amounts.

Sentinel is the score that watches for illness. Inside the confound window it keeps only a quarter of any skin temperature deviation, because temperature there is largely cycle-explained. It keeps half of an HRV deviation, because that one is partly explained. It keeps all of your resting heart rate and all of your breathing rate, untouched. Those two are the strongest tells of a real infection, so a genuine illness still surfaces during your luteal phase. Readiness does the matching thing, halving the pull an HRV dip has on your ceiling for the day.

Log a basal body temperature and the estimate gets firmer. When the luteal readings sit at least 0.2 degrees above the follicular ones, the phase is confirmed from your own data rather than from a calendar average. Apple Health sends those readings in on its own if you record them there.

A coach that cannot tell a luteal phase from a cold is guessing about half of its athletes.

The thing we are actually watching for

Cycle awareness earns its place for a harder reason than training smarter in a given week. It catches something that hurts people.

A period that arrives late, or stops arriving, is often the first visible sign of low energy availability. Under-fuelling that has gone on long enough to switch off a hormonal system has also had time to work on bone, and bone stress injuries are where that story usually ends. Female sex, low energy availability and menstrual disruption sit together in our own knowledge base as bone stress drivers.

So two signals get treated as early warnings. An overdue or absent period is one. A low ferritin value, whenever you have one from a blood test, is the other. Iron deficiency is common in endurance runners and more common again in menstruating ones.

When either is live, two numbers in the plan generator move. The peak volume multiplier drops from 1.6 to 1.4. The weekly ramp cap drops from 10 percent to 8 percent. Your proposed plan gets a lower ceiling and a gentler climb toward it, and it stays there until the signal clears. That adjustment only ever softens a plan. There is no path through that code that makes anyone’s training more aggressive.

Trailwise raises it with you in plain language, once and gently. It checks that you are eating enough for what you are doing. If it persists, it tells you to speak to a doctor or a registered dietitian. It does not name a condition, because it is not qualified to and neither are we.

Life stages are a coaching input, not a pause button

Two more inputs sit in your settings, under coaching boundaries, and they behave very differently from each other on purpose.

Pregnant or postpartum is the restrictive one. It turns off race building entirely. Trailwise will not prescribe intensity and will not give you a return to running date, because that date belongs to your clinician and to nobody else. What it does instead is coach a gradual return patiently, keep pelvic floor recovery in view, leave generous room for recovery, and treat consistency as the thing worth praising.

Perimenopause and menopause is a modifier. It is a life stage to train well through. Quality work stays available whenever the day’s signals support it, and the plan reads the day rather than capping your effort by default. One thing does change automatically: weekly strength sessions go from two to three, at the upper end of the evidence range, for bone and muscle. That one was a say-do gap we found and closed. The coach voice and the settings copy had been promising strength emphasis while the plan generator scheduled identical strength for everybody. Symptoms and medical questions, HRT included, go to your doctor.

The line we decided not to cross

The popular version of cycle-aware training is phase periodization. Hard weeks in the follicular phase, easy weeks in the late luteal phase, the whole block laid out against a predicted cycle. It is a good story and we do not do it.

Two reasons, and we wrote them down as a product decision rather than leaving them to whoever edits the code next.

The evidence that placing load by phase improves performance or reduces injury for recreational runners is thin and inconsistent. Building multi-week structure on a contested effect is a bad trade, because the athlete carries the cost and the structure is the hardest part of a plan to unwind.

And a calendar phase estimate carries real uncertainty. Committing a fixed block to predicted phases bakes that uncertainty into the skeleton. Correcting each day as it actually arrives holds up better, and it degrades gracefully when your cycle does something unexpected.

Adjust each day to the cycle. Leave the shape of the plan alone.

If replicated evidence lands, this changes. The plan skeleton already takes per-week emphasis inputs, so the hook is sitting there. What would move us is that evidence plus a phase estimate reliable enough to schedule against, which in practice means consistently temperature confirmed.

What it still cannot do

A femtech claim with no admitted gaps is worth very little, so here are ours.

Cycle data does not arrive from Garmin, Oura, Whoop or Fitbit. It comes from your check-in, or from Apple Health. That is the single most common thing people expect to work and it does not work yet.

Without a basal temperature the phase estimate is a calendar model. It says so, it carries a confidence, and it goes quiet and returns unknown rather than guessing when what you have logged is too thin or too old to place you.

If you use contraception that stops your period, the calendar model has nothing to work from. We have no good answer for that case today.

And all of it is opt-in. The cycle strip on your dashboard, the cycle notes in chat and the cycle side of your plan stay off until you switch them on. Ticking a period start is not a setting you can trip over by accident, and none of it happens quietly in the background before you have asked for it.

That is what the word means here. Being a woman is an input to the scores and to the plan, with real numbers behind it, an explicit line we do not cross, and a list of things that still do not work.

Trailwise is not a medical device. Neither is it a replacement for a doctor or a physiotherapist. Trailwise flags what is worth a conversation. It does not diagnose, and it will tell you when it does not know.