How the overnight reading works

Locate the night

Coach uses recorded sleep to choose a 4-hour window starting 30 minutes after sleep onset.

Summarize the readings

Heart-rate samples are averaged in 5-minute groups, then those groups are averaged for the night. This keeps bursts of samples from dominating the reading.

Compare with your baseline

The median from the last 7 days is compared with earlier readings within the last 60 days, excluding those recent days. The trend needs at least 3 readings in both periods, using the same source.

If data is missing

Coach can use a 1–5 a.m. window when sleep records are unavailable, then fall back to Apple Health resting-heart-rate readings if an overnight comparison cannot be made. A clock-based window does not confirm you were asleep.

Why it may differ from Health

Coach's overnight average is calculated from heart-rate samples. It is a different measurement from Health's resting-heart-rate value.

When it is available

Each usable night needs at least 12 5-minute groups containing samples. Without a heart-rate trend, at least 3 recorded feelings can still contribute to Fatigue.

How the signals combine

The recent heart-rate median uses 7 days and compares it with earlier readings within 60 days, requiring at least 3 readings in each period and flagging strain at 4% above baseline, or 2 bpm, whichever is more. Feelings need at least 3 answers, with a majority tired or sore counting as strain. HRV needs at least 3 readings in the last 7 days, and in at least 2 of the 4 weeks before. A weekly average 10% or more below the earlier weekly average counts as strain. Training monotony can corroborate when load data is usable and volume is not low, but cannot establish Fatigue alone.