virattt/ai-hedge-fund

An AI Hedge Fund Team

What it solves

This project provides a proof-of-concept framework for exploring how AI can be used to make trading decisions. It allows users to simulate trading strategies using AI agents without risking real capital, serving as an educational tool for AI-driven financial analysis.

How it works

The system uses a group of investor agents powered by various LLM providers (such as Anthropic, OpenAI, DeepSeek, and Google) to analyze financial data. It integrates with the Financial Datasets API to retrieve prices, fundamentals, and earnings data. Users can operate the system in two modes:

  • Paper trading: Simulates real-time market days with a ledger of fake money to track performance (NAV).
  • Backtesting: Replays a fund's strategy over historical data to compare its equity curve against a benchmark.

Who it’s for

It is designed for researchers and learners interested in AI-powered financial trading and the application of LLMs to market analysis.

Highlights

  • Multi-LLM Support: Compatible with a wide range of model providers including xAI, Kimi, and TypeSafe.
  • Hash-chained Ledger: Maintains a permanent track record of paper trading sessions in a local ledger.
  • Interactive Terminal App: Provides a direct interface for managing funds, executing sessions, and viewing history.
  • Integrated Backtesting: Allows users to test mandates against historical market data.

Related

  • Project
    TauricResearch/TradingAgents

    A multi-agent LLM framework that simulates a professional trading firm's workflow to analyze market conditions and inform trading decisions.

  • Project
    YUHAI0/fin-agent

    An AI-powered financial analysis assistant that uses LLMs and Tushare data to enable natural language stock screening, strategy backtesting, and portfolio monitoring.

  • Project
    Lumiwealth/lumibot

    A Python trading framework that enables the creation and deployment of AI agents and rules-based strategies across multiple brokers and asset classes.

  • Project
    lzwme/finance-quant-skills

    A collection of specialized AI agent skills for quantitative finance, enabling LLMs to handle financial data APIs, backtesting frameworks, and trading strategy development.

  • Project
    simonlin1212/TradingAgents-astock

    A fork of the TradingAgents multi‑agent research framework, specialized for China’s A‑share market. It adds three China‑specific analyst roles (policy, hot‑money, lock‑up), uses free Chinese data sources, and adapts trading rules (T+1, price limits, lot sizes). The pipeline (analyst → bull/bear debate → research manager → trader → risk debate → portfolio manager) is driven by configurable LLMs (MiniMax, DeepSeek, OpenAI, etc.) and can be run via a Streamlit web UI or a CLI. Includes performance‑statistics tooling, but is research‑only and not an investment service.