"JFKELLY89 team on Alpaca AI Trading Agents Hackathon Hackathon"

Team Idea

We build agentic trading systems where AI proposes and deterministic code decides. Our hackathon entry, Alpha Council, is an options-native paper-trading desk with a governance chain modeled on a real trading floor: a quant funnel finds candidates, GPT analysts and a portfolio manager debate them against Python-priced scenario payoffs, a Claude Red Team attacks every proposal, and a hard-coded Risk Constitution has the final word. The differentiator is measurement — every decision is stored three ways (what GPT proposed, what Claude changed, what executed), so each layer's P&L contribution is a number, and refused trades are audited too. Built on Alpaca's Trading API, market data, and MCP server. Links Code: https://github.com/JFKELLY89/alpha-council Demo: https://youtu.be/z-F0h_6jOp0

Team Members

John Kelly

john_kelly396

Submission

Alpha Council

An autonomous options paper-trading desk that debates its trades: a quant funnel finds candidates, GPT analysts propose, a Claude Red Team attacks, and a deterministic Risk Constitution decides — then measures whether each layer added value.