Notes on fatigue prediction for Kraft

Modeling athlete fatigue is mostly a data-quality problem, not a modeling problem. What I learned wiring training load and wellness signals into something coaches trust.

This is the studio bitácora — a working log rather than a polished publication. Entries get written when something is worth writing down: a strategy that cleared its evidence gate, a drawdown that taught me something, or an engineering decision I want to remember the reasoning behind.

The short version

Build small, ship it, keep it running, and write down what actually happened — including the parts that did not work. The full entry lives in the source repository; this page is the redesigned shell it renders into.

$ git log --oneline -1
a1b2c3d  notes on fatigue prediction for ...

Want the rest? It is part of the open notes at hello@namilabs.ai.