OpenInvest
Research-grade investment decision engine for AI agents: isolated multi-agent committee, auditable verdicts, backtests with lookahead protection, published nega…
- Transport
- Not stated
- Package
- —
- Registry id
- —
No install snippet on purpose. A working MCP config is a command, its arguments and an environment block — the last two are where API keys live, so this catalogue never stores them and cannot publish them. Follow the link above for the authors' own instructions.
OpenInvest four LLM roles cross-examine each other and a CIO aggregates one verdict. TradingAgents-style analyst agents scored below the naive baseline on 30-day directional hit rate. Calibration uses small-sample shrinkage with k=80 and band expansion with γ=1.1.
The system has dual execution paths that run the same prompts through two distinct implementations. The system uses YAML frontmatter for schema validation and Markdown body for LLM ingestion, with fcntl file locks for atomic transactions. The system never auto-trades; its internal ledger is a local record connected to no real payment or brokerage.
Return significance is bounded by t ≈ SR·√T, making calibration testable now.
Summary drafted from the project's own website. Every sentence is backed by text on that page and was reviewed before publishing.