ai-hedge-fund-net is a .NET alghoritmic program that provides trading signals by analyzing stocks using multiple AI agents. Each agent applies a different investment philosophy to decide whether a stock is a buy, hold, or sell. Agents also provide their reasoning, confidence score, key metrics, and specific rules behind each decision.

Currently, the following agents are implemented:

• charlie_munger (quality + management judgment)
• stanley_druckenmiller (macro, momentum, sentiment)
• ben_graham (deep value, margin of safety, balance sheet strength)
• cathie_wood (innovation, tech disruption)
• bill_ackman (activist investing, risk arbitrage)
• warren_buffett (value investing, moat, long-term)

Each agent integrates with an LLM (Large Language Model) trained for financial reasoning to generate the insights behind its signals.

This is an example of output for NVidia using warren_buffett and cathie_wood agents

1. Go to the Releases section and download the latest release.
2. Run the program with --help or -h to see usage instructions:

> AiHedgeFund.Console --help

3. Run the program by specifying one or more agents and one or more stock tickers:

> AiHedgeFund.Console --agent cathie_wood ben_graham --tickers MSFT AAPL

This project uses Alpha Vantage as the financial data provider. I have no affiliation or sponsorship with them—it simply happened that I created a free API key there and stuck with it. Once the initial porting and development phase is complete, I plan to support additional providers by implementing the IDataReader interface.

To use Alpha Vantage:

1. Get your free API key from: https://www.alphavantage.co/support/#api-key
2. Add the key to your appsettings.json file.

For LLM-based reasoning, OpenAI is used:

1. Get your API key from: https://platform.openai.com/account/api-keys
2. Add it to appsettings.json.

Example appsettings.json:

{
  "OpenAI": {
    "ApiKey": "your-openai-api-key"
  },
  "AlphaVantage": {
    "ApiKey": "your-alpha-vantage-api-key"
  }
}

All financial data is fetched from the remote API once, then serialized to disk and cached in memory. On subsequent runs with the same tickers, the program will use the cached data instead of re-fetching it.

To force a data refresh, manually delete the data folder located in the same directory as the program. A command-line parameter to automate this will be added soon.

This .NET project is loosely inspired by the ai-hedge-fund project written in Python.