AI-native CLI for trading Kalshi prediction markets with deep research and a 5-gate risk engine.
Run fundamental research, calculate model edge, and execute trades with half-Kelly sizing.
kalshi-trading-bot-cli is an AI-native command-line interface designed to find edge and execute trades on Kalshi prediction markets. It is built for algorithmic traders and quantitative professionals who want automated probability estimates and systematic execution.
The tool conducts deep fundamental research, generates independent probability estimates, and calculates edge by comparing model prices to live order books. Sigs are sized using half-Kelly and evaluated through a 5-gate risk engine before orders are placed.
It suits quantitative traders and developers familiar with the command line who trade prediction markets. It does not suit casual users looking for a graphical interface or non-technical traders without API access.
Generates independent probability estimates, ranked price drivers, and catalyst calendars for every market.
Filters trade signals through Kelly, liquidity, correlation, concentration, and drawdown checks.
Connects to external research APIs for AI-generated probability estimates and market-with-edge screens.
Builds diversified baskets, runs backtests, calculates correlation matrices, and validates portfolio parameters.
Turns settled positions into terse lessons and tracks Brier scores and category skill against the market.
Free under its open-source license, but requires paid API keys for Kalshi trading access and external AI research providers like Octagon and Tavily.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
These reviewed AI tools do the same job instantly — from our scored directory.
No, the bot explicitly requires Bun 1.1+ because it uses bun:sqlite and executes .tsx files directly.
All local configuration, caches, and the SQLite database are kept in the ~/.kalshi-bot/ directory.
Signals are sized using fractional half-Kelly sizing and filtered through a 5-gate risk engine.
See Zekai first in Google