Menstrual Cycle Phases: How to Read Training and Recovery in Context

The Four Phases of Your Cycle, and What Drives Each One

A typical cycle runs about 28 days, though anywhere from roughly 21 to 35 days is considered normal, and length varies from person to person and cycle to cycle. Four phases make up that cycle, and each one is defined by which hormone is doing the driving, not by a fixed number of days.

Janse de Jonge's widely cited review of menstrual cycle research explains why phase boundaries are approximate rather than fixed: hormone concentrations fluctuate in a pulsatile way, cycle length itself varies, and the interaction between estrogen and progesterone changes across the month rather than switching on and off at a clean line. Treat the day ranges above as a starting orientation, then let your own temperature and symptom data refine where your phase boundaries actually fall.

How Your Wearable Data Actually Shifts by Phase

Large studies that pull data directly from wearables, rather than small lab samples, give a clearer picture of which signals move predictably across the cycle and which do not. A 2024 study by Alzueta and colleagues, published in the Journal of Biological Rhythms, tracked finger temperature, heart rate, and sleep across full cycles in young and midlife women using wearable and diary data. Temperature and heart rate showed a consistent cycle-related pattern. Sleep metrics did not show the same consistent variation in that dataset, which is a useful check against assuming every recovery input moves in lockstep with hormones.

A larger 2026 analysis by O'Day and colleagues in npj Digital Medicine, built from over a million days of wearable data across thousands of women, reinforced that cardiorespiratory metrics vary across the cycle, and that how much they vary depends partly on a person's typical cycle length. It also found that short-term sleep loss shifts resting heart rate regardless of cycle phase. That is a useful reminder that phase is one input into your recovery data, not the only one, and it is worth reading your recovery score alongside your HRV and sleep data rather than any single number in isolation.

Reading Your Recovery Score in Context, Not as an Alarm

The most common mistake with cycle-aware data is treating an expected luteal-phase dip as a warning sign. The second most common mistake is the opposite: ignoring a real drop in readiness because it happens to land in the luteal phase.

Using Phase Context for Training Without Overcorrecting

It is tempting to treat cycle phase as a strict training rulebook: hard days only in the follicular phase, easy days only in the luteal phase. The research does not support that level of rigidity. McNulty and colleagues' 2020 systematic review and meta-analysis in Sports Medicine, which pooled results across dozens of studies, found only a trivially small average reduction in exercise performance during the early follicular phase compared to the rest of the cycle. The review's own conclusion was that the effect size is small, the variation between studies is large, and general phase-based performance guidelines are not well supported by the current evidence. A personalized approach, built from your own data over multiple cycles, is more useful than a blanket rule.

Symptoms track more closely with training disruption than phase alone does. Bruinvels and colleagues surveyed 6,812 exercising women through the Strava app and found that mood changes and anxiety, tiredness and fatigue, stomach cramps, and breast pain or tenderness were the most commonly reported symptoms, and that symptom frequency was associated with reduced availability to train and compete. That points to a more useful practice than reading the calendar: log how you actually feel alongside your wearable data, and use a two-week strength and intensity framework built around the follicular and luteal phases as a starting template rather than a fixed rule.

Where the Data Is More Settled

Not every cycle effect is trivial. Chidi-Ogbolu and Baar's review of estrogen's effect on musculoskeletal tissue found that estrogen supports muscle mass, strength, and collagen content in tendons and ligaments, but also reduces tendon and ligament stiffness at high estrogen levels. That combination is part of why the brief estrogen peak around ovulation is associated with a small, well-documented rise in ACL and ligament injury risk during high-speed cutting and jumping movements. That is a narrow window of caution, not a reason to change training for the whole month.

The Biggest Misconception About Cycle Phase and Performance

The evidence points the other way. The average performance difference between phases is small enough that a rigid phase-based schedule will often be wrong for a given person on a given day. What is well documented is that symptoms, sleep, and training load interact with phase, and that wearable temperature and heart rate data track phase transitions fairly reliably even when performance itself does not move much. The practical use of phase tracking is context, not prediction: it explains why a luteal-phase HRV reading looks lower than your follicular one, it flags when a temperature or heart rate shift is expected rather than concerning, and it gives you a consistent lens for interpreting your own multi-cycle data. It is not a substitute for reading how you actually feel and perform on a given day.

Frequently Asked Questions