How AI Hedge Fund Works

AI Hedge Fund simulates an AI-powered hedge fund using 18 different investment strategy agents. Here's a complete guide to understanding the process from start to finish.

🚀 Try It Now 📖 Complete User Guide

🔄 The Complete Process

AI Hedge Fund follows a structured workflow that simulates how a real hedge fund operates, but with AI agents making the decisions. Here's how it works:

1

Select Stocks

Choose any stock tickers you want to analyze (e.g., AAPL, MSFT, NVDA, TSLA). Free tickers include major tech stocks. For other stocks, you'll need a Financial Datasets API key.

Example:
  • Free tickers: AAPL, GOOGL, MSFT, NVDA, TSLA
  • Multiple stocks: Analyze several companies simultaneously
  • Compare sectors: Different industries or markets

You can analyze multiple stocks simultaneously to compare different companies or sectors.

2

AI Agents Analyze

Our 18 AI agents analyze the stocks using their unique strategies. Each agent examines the companies from their perspective and generates trading signals (BUY, SELL, HOLD).

What Each Agent Does:
  • Warren Buffett: Looks for wonderful companies at fair prices
  • Peter Lynch: Seeks growth companies you can understand
  • Technicals Agent: Analyzes chart patterns and indicators
  • Sentiment Agent: Evaluates market sentiment and news

Each agent provides detailed reasoning for their decisions, helping you understand their investment philosophy. You can see their analysis in real-time as they process information.

3

Risk Manager Evaluates

The Risk Manager evaluates all signals from the analyst agents. It calculates risk metrics, sets position limits, and ensures the portfolio doesn't take excessive risk.

Risk Management Functions:
  • Calculates Value at Risk (VaR)
  • Sets maximum position sizes
  • Enforces diversification rules
  • Evaluates correlation between positions
  • Monitors portfolio risk metrics

This step is crucial for protecting your portfolio from large losses. The Risk Manager acts as a safety net, ensuring no single position or strategy can cause excessive risk.

4

Portfolio Manager Decides

The Portfolio Manager makes final trading decisions based on all agent recommendations and risk constraints. It determines position sizes, executes trades, and manages the portfolio.

Portfolio Manager Functions:
  • Aggregates signals from all agents
  • Determines optimal position sizes
  • Executes buy/sell decisions
  • Manages cash and positions
  • Rebalances portfolio as needed

You see the final portfolio allocation and trading decisions. The Portfolio Manager considers all agent recommendations and risk limits to make optimal decisions.

📊 Trading Flow Structure

A typical trading flow follows this pattern:

Start ├─ Tickers Node (AAPL, MSFT, NVDA) ├─ Analyst Agents (parallel) │ ├─ Warren Buffett Agent │ ├─ Peter Lynch Agent │ ├─ Technicals Agent │ └─ Sentiment Agent │ ↓ Risk Manager (aggregates signals, calculates risk) ↓ Portfolio Manager (final decisions, executes trades) ↓ End (results displayed)

Learn more about building flows and available agents.

💡 Example Trading Flows

Basic Flow Example

Goal: Analyze AAPL and MSFT using value and growth strategies

  1. Add Tickers Node: Set tickers to AAPL, MSFT
  2. Add Analyst Agents: Warren Buffett, Peter Lynch, Technicals Agent
  3. Connect Nodes: Tickers → All agents → Risk Manager → Portfolio Manager
  4. Configure: Set initial cash ($100,000), margin requirement (50%), select LLM model
  5. Run: Click "Run" and watch the analysis progress
Expected Results:
  • Each agent provides BUY/SELL/HOLD signals
  • Risk Manager evaluates and sets limits
  • Portfolio Manager makes final decisions
  • See portfolio allocation and reasoning

See our complete guide for detailed instructions.

Multi-Strategy Analysis

Run multiple analyst agents in parallel, then aggregate their signals through the Risk Manager before final portfolio decisions.

Multiple Analyst Agents (parallel) ├─ Warren Buffett ├─ Peter Lynch ├─ Technicals Agent └─ Sentiment Agent ↓ Risk Manager (aggregates signals) ↓ Portfolio Manager (final decisions)

Use Case: Get diverse perspectives on the same stocks. Each agent brings a different analytical approach, providing comprehensive coverage.

Sequential Analysis

Chain agents together where one agent's output feeds into the next. For example, fundamentals analysis feeds into valuation.

Fundamentals Agent ↓ Valuation Agent (uses fundamentals) ↓ Risk Manager ↓ Portfolio Manager

Use Case: Build complex analysis pipelines where each step builds on the previous one. This allows for deeper, more nuanced analysis.

🎯 Common Strategy Patterns

Combine agents in these proven patterns for different investment approaches:

Conservative Strategy

Agents: Warren Buffett + Ben Graham + Risk Manager

Focus on value investing with strong risk management. Ideal for conservative investors seeking stable returns and capital preservation. Emphasizes companies with strong fundamentals and competitive moats.

View Value Agents →

Growth Strategy

Agents: Peter Lynch + Cathie Wood + Risk Manager

Focus on high-growth companies with innovation potential. Perfect for investors seeking above-average returns through growth. Targets companies that can scale rapidly and transform industries.

View Growth Agents →

Balanced Strategy

Agents: Multiple value + growth agents + Risk Manager

Combine different approaches for diversified analysis. Provides comprehensive market perspective by balancing value and growth considerations. Reduces reliance on any single strategy.

View All Agents →

Technical Strategy

Agents: Technicals Agent + Sentiment Agent + Risk Manager

Focus on technical indicators and market sentiment. Good for short to medium-term trading strategies. Relies on chart patterns, indicators, and market psychology.

View Analytical Agents →

Fundamental Strategy

Agents: Fundamentals Agent + Valuation Agent + Risk Manager

Deep fundamental analysis with valuation. Ideal for long-term value investors. Focuses on understanding company financials and determining intrinsic value.

View Analytical Agents →

Multi-Perspective

Agents: Value + Growth + Technicals + Risk Manager

Comprehensive analysis from multiple angles. Best for thorough stock evaluation. Combines fundamental, technical, and sentiment analysis for complete picture.

View All Agents →

📈 Backtesting

Backtesting lets you test strategies on historical data to see how they would have performed.

Setting Up a Backtest

  1. Build your flow (same as regular analysis)
  2. Set date range (start and end dates)
  3. Set initial capital (e.g., $100,000)
  4. Click "Backtest"

Backtest Results Include:

  • Daily portfolio value
  • Total return (percentage)
  • Sharpe ratio (risk-adjusted returns)
  • Max drawdown (largest peak-to-trough decline)
  • Win rate (percentage of profitable trades)
  • Final positions and cash
💡 Best Practices:
  • Test on different time periods
  • Use realistic transaction costs
  • Consider market conditions
  • Compare multiple strategies
  • Review individual trade decisions

Read more about backtesting best practices.

🖥️ Using the Web Interface

The web interface provides a visual way to build and run trading flows.

Left Sidebar

Agent library and flow management. Drag agents onto the canvas.

  • Browse all 18 agents
  • Save and load flows
  • Access templates

Center Canvas

Visual flow builder. Drag-and-drop agents and connect them with edges.

  • Drag agents from sidebar
  • Connect nodes with arrows
  • Zoom and pan for large flows

Right Panel

Node configuration. Set tickers, models, and parameters for each agent.

  • Configure selected nodes
  • Set stock tickers
  • Choose LLM models
  • Adjust parameters

Bottom Panel

Execution results and logs. See real-time analysis progress and results.

  • Real-time progress updates
  • Agent reasoning and signals
  • Portfolio decisions
  • Error messages and logs

Learn more about the web interface features.

Ready to Get Started?

Now that you understand how it works, start building your first trading flow.

🚀 Access Application Now
Complete User Guide Full Documentation Explore Agents