BLOG

Research & Strategy

Deep dives on weather market mechanics, forecast blending, and risk management for algorithmic trading.

Subscribe via RSS

Introducing Weatherman Trading: A Sovereign Quant System

A self-hosted automated trading engine for Kalshi weather markets, with transparent risk management. We set it up on your server and train you to run it.

algorithmic tradingweather marketsKalshiquantsource code

Weather markets are real. The infrastructure for automated trading on them doesn't have to be a black box.


Weatherman Trading is a dual-daemon execution engine that runs on your own server, trades live Kalshi weather contracts on a five-minute loop, and gives you full access to every line of Python source code.


The Problem We Solved


Most trading systems sold today ask you to trust a black box:

  • You don't see the code

  • You don't control the infrastructure

  • You can't audit the risk gates

  • You pay a subscription and hope it works

  • Weather markets are different. The edge comes from correctly modeling the atmosphere, not from information asymmetry between traders. Someone else running the same forecast blend on the same contracts doesn't erode your position — the underlying temperature doesn't change because more people are watching the same models.


    What we're selling is the engineering: the daemon architecture, the safety gating, and the sizing math, fully readable and modifiable.


    What's Inside


  • Execution daemon: Runs the trading loop on a fixed five-minute cadence

  • Telegram operator daemon: Receive status updates, pull balances, trigger kills without touching the trading loop

  • GFS/ECMWF forecast blending: Dual independent forecast models feeding every decision

  • Kelly-fraction sizing: Position size computed as a capped fraction of the mathematical Kelly optimum

  • L2 liquidity capping: Orders sized to the actual depth available at the top of the book

  • Regional exposure netting: Opposing bets within the same region offset in capital math

  • WAL-mode SQLite ledger: Every trade, quote, and position written with write-ahead logging so concurrent writes don't lock

  • Run it on your own VPS. You own the keys, you own the data, you own the logs.


    Why We Built This


    We spent three years trading weather markets manually. We built automation to remove the discretion, and it worked — but only because we could read the code, understand the assumptions, and modify it when the market changed.


    A locked black box would have failed the moment we needed to adjust for a new forecast provider or a rule change on the exchange.


    So we decided to sell the engineering instead.


    Next Steps


    We install the engine, its deployment stack and the phone dashboard on your own server, connect your Kalshi account, and train you to run it: Setup + Training is $1,500 one-time, or we can run it for you on the Managed plan. See the pricing section, and our live account on the results page.


    Start with the architecture guide to understand the seven-stage pipeline. Then read the math section to see how fair value, volatility, exposure, and bet size all reduce to fixed formulas.


    Your bot will fail safe, not fail silent.