TriPeak

Fitness, fatigue and form

Three curves drawn from one input. Every session contributes its TSS, and the difference between how fast the two averages respond is the whole idea.

The three numbers

This is the Banister impulse-response model: one training impulse produces both a fitness effect and a fatigue effect, and they decay at different rates. Because fatigue decays roughly six times faster than fitness, a period of rest drops ATL quickly while CTL barely moves — and the gap between them opens up. That gap is form, and producing it deliberately is what a taper is.

CTL — fitness

An exponentially weighted average of daily TSS with a 42-day time constant. Slow to build, slow to lose.

ATL — fatigue

The same average with a 7-day constant. Rises and falls quickly.

TSB — form

Yesterday's CTL minus yesterday's ATL. Positive means fresher than your recent load; negative means buried in it.

What form is worth on race day

A TSB somewhere around +10 to +20 on race morning is the window that most often produces a good race. Below it and you arrive carrying fatigue; far above it and you have rested long enough to have lost sharpness, usually along with some fitness.

The window is a guide, not a law. Athletes differ, and a TSB of +25 on a high CTL is a different animal from +25 on a low one.

How to read a ramp

The rate CTL climbs matters more than the value it reaches. Adding more than about five to seven CTL points a week is the region where injury and illness start appearing, and the number does not care that you felt fine at the time.

The ways the chart lies

• It is only as good as the TSS feeding it, and TSS is only as good as the thresholds. A stale FTP inflates every point on the curve.
• CTL has no memory of what the training was. A hundred points a day of easy riding and a hundred points a day of intervals build the same curve and not the same athlete.
• Missing data reads as rest. A week of unsynced sessions looks exactly like a week off, and the chart will cheerfully tell you that you are fresh.
• It knows nothing about sleep, work, illness or heat, all of which move how a given TSB actually feels.
• Early numbers are unreliable. The averages need roughly six weeks of history before CTL means very much.