How it works
Everything on your dashboard — your cycle ring, your phase, the timing behind your plan — comes from a handful of calculations. This page explains what they are, in plain language and pictures, and what evidence sits behind each one.
For the research behind Liv's actual recommendations — the studies, the sample sizes, the confidence ratings — see Liv's Brain. This page is about the maths, not the coaching.
Liv starts from what's generally true, then quietly hands over to what's true for you, the moment she has enough of your own data to trust it more than the average.
Today's log, if you've made one. Nothing beats what you actually told her this morning.
Your own history, if you have some. If you haven't logged today, she looks at how you've felt at this same point in previous cycles.
What research generally finds, as a starting point, until your own data can take over.
Your cycle is split into four stretches — menstrual, follicular, ovulation, luteal — sized from two things: the date your last period started, and how long your cycles usually run.
Day 8 of ~28
This is the same ring you see on your dashboard. It always starts at day 1 (12 o'clock) and runs clockwise through the four phases, sized to how long each one lasts for this cycle — not a generic clock face.
The one deliberate choice worth explaining is where ovulation gets placed. It isn't a fixed percentage of your cycle. It's anchored backwards from your next expected period — about 14 days before it — because that back half of the cycle varies far less than the front half does, from person to person and cycle to cycle. What actually stretches or shrinks a longer or shorter cycle is almost entirely the first half — the run-up to ovulation, not the run-down from it. Below is that same idea across three different cycle lengths:
Notice the dashed luteal segment stays close to the same width in all three — it's mostly the days before it that expand or contract.
Why anchor it this way: after ovulation, the follicle that released the egg becomes a short-lived structure (the corpus luteum) that reliably breaks down after about two weeks if there's no pregnancy — which is what triggers your next period. That “two weeks and done” mechanism is why the second half of the cycle is far more consistent than the lead-up to ovulation, which genuinely differs cycle to cycle. Three separate cohort studies — two very large, one smaller but tracking the same women repeatedly over a year — all find the same pattern: follicular length is what actually drives cycle-length variability, while luteal length moves comparatively little.
It's worth being precise, though: the most rigorous of the three found the luteal phase is not perfectly fixed — just far steadier than the follicular phase. “Always exactly 14 days” oversells it; “the far more consistent half of your cycle” is the honest version, and the one this app builds on. See the three studies in Liv's Brain →
Once you've logged a few bleed days, Liv uses your actual logged period length for the menstrual stretch — not an estimate. She groups your logged bleed days into a single period even if there's a light day or a gap in the middle, and only starts a new period once there's a real gap between bleeding.
A gap of more than 10 days between logged bleed days always starts a new period, even if a day or two inside a period goes unlogged.
Before you've logged any bleed days at all, she starts with a placeholder: roughly the first 17% of your average cycle length.
This isn't fixed to whatever you entered when you signed up. It's a rolling average of your last three logged cycles, and it keeps updating as you log more.
Only the three most recent reasonable gaps go into the average. A confirmed skipped cycle — however long — is left out entirely, so it can't drag your “usual” length off course.
Gaps that look like data mistakes (under 15 days or over 90) are ignored the same way, without needing you to confirm anything.
Every phase Liv shows you carries one of two confidence levels, even if it isn't always spelled out on screen:
Tracked means today falls on or after a period you've actually logged. Estimated means it's built from your average cycle length and last known start date instead, because there's no closer data point yet.
If you've told us you have PCOS and an irregular cycle, Liv doesn't attempt to place you in a menstrual/follicular/ovulation/luteal phase — the timing genuinely isn't predictable enough for that to be honest. Instead she switches to a mode built around what tends to help regardless of timing: steady movement, blood sugar-friendly meals, and asking how you're feeling rather than assuming it.
One honest gap today: if you're on hormonal contraception, Liv still estimates a menstrual-style cycle for you, even though ovulation is often suppressed by the method you're using. That's a known limitation we want to fix, not a design choice.
Cycle-phase research doesn't all carry the same weight, so Liv's Brain grades every recommendation before it's ever handed to you:
The stronger the evidence, the more confidently Liv states something. Where it's thin, she hedges — on purpose.
Most of the research this app draws on was done on young, white, regularly-cycling women without health conditions — university students and recreational athletes, mostly. If that isn't you, the population averages may simply apply less, and that's exactly why the ladder at the top of this page exists: your own logged data is allowed to outrank the general research, on purpose, the moment there's enough of it.
A few specific areas are still thin across the whole field, not just this app: how cognition really shifts across the follicular phase, whether there's a genuine performance “peak” at ovulation (the popular version of this claim outruns what the data actually shows), and cycle-specific protein needs. Liv is built to hedge on these rather than state them as settled.
This page describes the calculations as they exist in the product today. It does not yet cover hormonal contraception handling — that's a known, unfixed gap, not an oversight in this page.
