Multi-agent investing, explained step by step
This project runs simulated trades only. Use it to compare agent teams, prompts, model choices, and decision quality across stocks, FX pairs, ETFs, crypto, or any symbol you want to test with a manual reference price.
What to do next
Use an account seeded into this project's SQLite database.
Choose model provider, model name, rounds, and default role prompts.
Enter a stock, FX pair, ETF, crypto symbol, or other paper-traded asset.
Review the decision and apply it with a reference price.
Leaderboard tracks prompt and Agent Set versions, not just users.
Rankings compare agent configurations, not just users
Every Agent Set version keeps its own performance history, so changing a prompt or model starts a new comparable version instead of rewriting past results.
No ranking data yet. Run a decision and apply it to paper trading to create a snapshot.