This is the judgment feezify encodes, in plain language. It’s open on purpose: the method is the point, not a secret. The numbers and the light are deterministic — computed by a fully unit-tested engine, not guessed by the AI.
This page mirrors the canonical
src/adapters/claude-skill/method.md, which wins in case of conflict. The exact constants are calibrated per athlete and may evolve.
Two readings, crossed
feezify reads two things and crosses them.
Objective — your form (TSB)
From your activities it rebuilds a daily training load (TSS), preferring power, then heart-rate, then session-effort (RPE) — always recomputed from raw when possible, so the series stays consistent. (Strava’s own “relative effort” is flagged incompatible and recomputed.) From that load it tracks:
- CTL (fitness) — a slow ~42-day exponential average of load.
- ATL (fatigue) — a fast ~7-day exponential average of load.
- TSB = CTL − ATL of yesterday (form).
TSB bands:
| TSB | Meaning |
|---|---|
| above +15 | fresh |
| +5 to −10 | optimal |
| −10 to −30 | productive load |
| below −30 | overreaching |
Subjective — your readiness
From your daily journal (sleep, fatigue, motivation, mood, stress, appetite, thirst) it reads a readiness score. Appetite and thirst are read as deviation from your own normal, in either direction — not “more is better”.
If you wrote no journal entry for a day, the score is a neutral default, and feezify says so plainly — it never presents an unmeasured score as measured.
The crossing rule (the wedge)
The subjective gates the objective. This is the most evidence-backed part of the whole system: when how you feel and what the numbers say disagree, how you feel wins.
- Injury or illness → red. Always. No number overrides it.
- Low readiness is never green, even on fresh legs.
- High readiness + non-negative form → green.
- High readiness + deeply negative form → amber (assume hidden fatigue) — unless you’re knowingly inside a build block, where that load is the plan.
- A rest day reframes a low score as expected, not alarming.
The “why” is always the lowest-scoring markers and what your own words say is driving them.
Why deterministic + AI, together
- The engine computes the numbers and the light. It’s pure, testable, and reproducible — the same inputs always give the same read. The AI is instructed not to recompute or override it.
- The AI reads your narrative journal across days — the patterns a spreadsheet can’t see — and writes the read in plain, honest language, tracing every claim back to a journal day.
That split is deliberate: math you can trust, judgment that stays human-readable.
The one-line summary
Load gives your form. Your journal gives your readiness. How you feel gates the numbers. feezify reads; you decide.
Back to → Daily use · Docs index