TreadLab (TireBot)

TreadLab (formerly TireBot) recommends a gravel race tire and front/rear pressures for a specific event route. It blends published rolling-resistance (CRR) data with your course model: segments tagged by surface (road, gravel categories, and optional “above category” rough sections), distance, and how hard the early race is weighted—because the first quarter often decides whether you stay with the front group.

The scoring engine combines route-weighted rolling loss, an aero width penalty (optionally scaled using GPX so steep climbs and descents count less toward that term), optional tire-mass cost over elevation, and a rough-surface impedance proxy so setups that are too stiff on chunky terrain pay a penalty beyond raw CRR tables. It outputs ranked tires, suggested pressures, alternatives, and an estimated power breakdown—plus a tire issue risk label (Low / Medium / High) from either heuristics or a small kNN model trained on your own labeled events.

There’s a Streamlit app and a CLI; the published whitepaper walks through inputs, math, limits, and safety notes (pressures are starting points, not a substitute for manufacturer limits or real-world validation).