AI & LLM-Driven Bots

BEXAI / KalshiBot

by BEX (BEXAI)

View on GitHub →

What it does

Dual-loop architecture: Google TimesFM 2.5 for fast quantitative tick predictions, Gemma 4 31B for slow multi-agent LangGraph debate (Bull, Bear, Volatility Forecaster, and Risk Management personas). Built without an SDK using manual RSA-PSS payload cryptography. React/Vite frontend exposes AI inference thresholds and live P&L.

What we learned from it

the dual-loop pattern (fast model + slow reasoning) is the most sophisticated architecture we've seen in open-source prediction-market bots. Trade only when both loops agree is a strong default.

Find the author

Related bots

Other open-source Kalshi bots in the same category:

OctagonAI / kalshi-deep-trading-bot

★ 206

The most-starred AI-driven Kalshi bot, period. Performs deep fundamental research per market via the Octagon Research API, generat...

ryanfrigo / kalshi-ai-trading-bot

Grok-4 integration with multi-agent decision-making and portfolio optimization. Uses 0.25 Kelly sizing with a category scorer that...

ajwann / kalshi-genai-trading-bot

"Vibe-coded" generative AI bot — system-prompts an LLM (Grok) into the role of a professional prediction-market trader. More exper...

Don't want to maintain your own bot?

Bot for Kalshi is the no-code platform for automated prediction-market trading — visual builder, signals, encrypted credentials, kill switch. The infrastructure these open-source projects build, ready to use.

Start Free Trial →