DK
Back to active records

PX-02

Prototype

Machine-learning experiment

Weather Markets.

A machine-learning experiment for Kalshi's daily weather markets.

Record / In development

Scope
Daily weather
Method
Model-driven
State
Active prototype

System path / 04 stages

  1. 01

    Question

    Define a narrow market hypothesis

  2. 02

    Model

    Test a weather-driven signal

  3. 03

    Market

    Compare the signal with available prices

  4. 04

    Review

    Measure outcomes and revise

Decision notes / 03

D-01

Keep the market narrow

The experiment is limited to daily weather contracts so model behavior and outcomes remain inspectable.

D-02

Treat it as research

The project is presented as an active prototype; implementation notes and results will be added as they stabilize.

D-03

Separate signal from outcome

The useful question is whether a model adds information, not whether an individual market resolves favorably.