Italianhype has released BLUM, an open-source project designed to explore how autonomous systems can research markets, generate trading hypotheses, and improve through structured feedback. The platform aims to combine technical analysis, sentiment data, and macroeconomic context within a multi-agent architecture.
- BLUM features a central financial brain orchestrating specialized agents for market scanning, thesis generation, and paper-trading simulation.
- The system includes a deterministic risk engine that no language model can bypass, ensuring safety in decision-making.
- It implements a continuous learning loop with structured financial memory to record trade outcomes, failures, and market conditions.
- The architecture supports a distributed "Shared Brain Network" where nodes contribute validated strategies and evidence without overwriting global models.
- Current research focuses on Forex paper trading, reinforcement learning, and integrations with projects like FinGPT, FinRL, and Qlib.
The project is positioned as a transparent, reproducible research platform for the financial-AI community rather than a tool guaranteeing profitability.