Summary
The creator behind All About AI walks through a personal system for running multiple independent AI-assisted trading strategies simultaneously — what he calls “pods” — using Claude Fable 5 for data analysis and Codex for building and querying data pipelines.
The video covers two live setups. The first is a Polymarket 5-minute market-maker strategy that had been running approximately 24 hours at the time of recording, generating 18 fills and roughly $76 in profit during an unusually quiet market session. The second is a mean-reversion pair trade built on correlated equities. Using Fable 5 to analyze five years of closing price data for Coca-Cola and PepsiCo, the model found that the pair’s historical correlation has broken down following GLP-1 headwinds for PepsiCo — making it a poor current candidate. Switching to a VMware/MA pair produced significantly stronger results: 15 winners out of 21 historical trades, with the strongest individual setup yielding approximately 4%.
The broader philosophy behind the pod system is portfolio-level thinking applied to automated strategies: accept that individual pods will lose money, keep them non-correlated, and avoid intervening emotionally in any single one — similar to holding an index fund. The creator compares over-managing a single pod to forcing trades rather than letting expected value play out. The workflow demonstrates a practical loop of using Codex to source financial data, Claude Fable 5 to analyze and backtest it, and then setting up autonomous execution once a strategy passes statistical review.
📺 Source: All About AI · Published June 15, 2026
🏷️ Format: Workflow Case Study
