AI Hedge Fund is a proof of concept that explores using LLM agents to make trading decisions. It is for educational and research purposes only: it places no real trades and gives no investment advice.
Investor agents run on a model from one of the supported providers, and Financial Datasets supplies prices, fundamentals, and earnings data. The interactive terminal app prompts for keys the first time it needs them.
The app has two modes. Paper trading runs your funds on real market days with fake money: each run shows the exact decision for approval before it executes, a kill switch halts a fund, and every session is appended to a hash-chained ledger so net asset value builds into a track record. Backtesting replays a fund over history and draws its equity curve against a benchmark. Fund definitions set strategies, capital, and cadence, and carry no tickers, so you choose the stock universe for each backtest.
Features
- Investor agents: LLM agents generate the trading decisions for each fund
- Paper trading: funds run on real market days with fake money and a per-run approval step
- Hash-chained ledger: each session appends to a tamper-evident record of the fund's history
- Kill switch: halts a paper fund
- Backtesting: replays a fund over history and plots its equity curve against a benchmark
- Fund definitions: strategies, capital, and cadence defined once, with tickers chosen per backtest
- Model providers: Anthropic, OpenAI, DeepSeek, Google, xAI, Kimi, and TypeSafe
