A plan is a hypothesis. The log is the result.

A training plan is written once, against a real distance and a real date. It stops being accurate the moment the first week is run. The method here is to keep rewriting it from what was logged, not from what was intended.

LOGGED

Every run arrives from Strava within a minute or two of being saved — distance, moving time, elevation gain, average pace, and heart rate where a watch recorded it. Nothing is entered twice, and nothing is estimated on the way in. A run that was never recorded stays absent rather than being filled in.

READ

Each planned session carries an intended load. Each logged run carries an actual one, built from duration and intensity together — an easy hour and a hard twenty minutes are not the same demand, and are not counted as though they were. Those actuals accumulate two ways: a slow average that represents fitness, and a fast one that represents fatigue. The gap between them is the whole signal.

The comparison is always against the plan the week was written against, never against a rolling average of everyone or of some previous season. A week can be well under a normal load and still be too much, if it was written to be a recovery week.

ADJUSTED

When the gap opens — a session run far harder than asked, a run missed, three days of climbing where the plan expected flat — the next session is rewritten before it arrives. Usually that means distance comes down, or intensity does, or a rest day moves. The change is stated in one line, in plain words, along with what caused it. There is no score, no readiness ring, and no number without a sentence attached.

Adjustments are conservative by design. A single hard session moves tomorrow; it does not rewrite the block. A pattern across a week does.

Currently trained on Strava activity data, synced from a watch or phone GPS. A direct Apple Health connection is planned. Nothing here reads a live wearable feed, and no part of the method depends on one.