CryptoTrader Documentation

Overview

CryptoTrader is an AI-driven cryptocurrency trading plan generator powered by crewAI. The system employs multiple AI agents that collaborate to analyze cryptocurrency market data, research news and sentiment, and develop actionable trading strategies.

Project Structure

crypto_trader/
├── output/            # Directory for storing generated trading plans
├── .venv/             # Python virtual environment
├── knowledge/         # Knowledge base for the project
├── src/               # Source code
│   └── crypto_trader/
│       ├── config/
│       │   ├── agents.yaml  # Agent configurations
│       │   └── tasks.yaml   # Task configurations
│       ├── tools/
│       │   └── binance_rapidapi_tool.py  # Tool for Binance API interactions
│       ├── crew.py    # Main crew definition with agents and tasks
│       ├── main.py    # Entry point and CLI interface
│       └── __init__.py
├── tests/             # Unit tests directory
├── README.md          # Project documentation
├── pyproject.toml     # Project configuration
└── .gitignore         # Git ignore file

Installation and Setup

Prerequisites

• Python 3.10 or higher (< 3.13)
• UV for dependency management

Installation Steps

1. Install UV:

pip install uv

2. Clone the repository and navigate to the project directory

3. Install dependencies:

crewai install

4. Configure API keys:
   • Add OPENAI_API_KEY to .env file (for LLM access)
   • Add RAPIDAPI_KEY and RAPIDAPI_BINANCE_HOST to .env file (for Binance data access)
   • Add SERPER_API_KEY to .env file (for web search capabilities)

Core Components

AI Agents

The system employs four specialized AI agents:

1. Binance Data Analyst

   • Role: Fetch and process cryptocurrency market data from Binance
   • Tools: BinanceRapidApiTool
   • LLM: nvidia_nim/meta/llama-3.3-70b-instruct

2. News and Sentiment Researcher

   • Role: Research and analyze news and market sentiment
   • Tools: SerperDevTool, ScrapeWebsiteTool
   • LLM: nvidia_nim/meta/llama-3.3-70b-instruct

3. Crypto Analyst

   • Role: Analyze market data and news to provide comprehensive insights
   • LLM: nvidia_nim/deepseek-ai/deepseek-r1

4. Trading Plan Strategist

   • Role: Develop actionable trading plans based on analysis
   • LLM: nvidia_nim/deepseek-ai/deepseek-r1

Tasks

The agents collaborate on a sequence of tasks:

1. Fetch Binance Klines Task: Retrieve historical candlestick data
2. Fetch Binance Price Ticker Task: Get current price information
3. Fetch Binance 24hr Stats Task: Get 24-hour market statistics
4. Research News and Sentiment Task: Gather and analyze recent news
5. Analyze Crypto Data Task: Synthesize market data and news
6. Develop Trading Plan Task: Create an actionable trading strategy

Tools

The project uses custom and third-party tools:

1. BinanceRapidApiTool: Custom tool for accessing Binance cryptocurrency data via RapidAPI
2. SerperDevTool: For web searching capabilities (from crewai_tools)
3. ScrapeWebsiteTool: For extracting information from web pages (from crewai_tools)

Usage

Basic Usage

Run the project with default configuration:

crewai run

This will:

1. Generate a trading plan for the default cryptocurrency (PEPEUSDT)
2. Save the report to the output/ directory

Advanced Usage

Customize the execution with command-line arguments:

python -m src.crypto_trader.main --symbol BTCUSDT --interval 1h --limit 100 --output custom_output --verbose

Available parameters:

• --symbol or -s: Cryptocurrency trading symbol (default: PEPEUSDT)
• --interval or -i: Kline interval (default: 4h)
• --limit or -l: Number of klines to fetch (default: 72)
• --output or -o: Output directory (default: output)
• --verbose or -v: Enable verbose logging

Output

The system generates a comprehensive trading plan in Markdown format containing:

1. Market data analysis (price, volume, trends)
2. News and sentiment analysis
3. Support and resistance levels
4. Entry and exit points
5. Risk management strategies
6. Profit targets

The output is saved to a file named trading_plan_{target_symbol}_{datetime}.md in the output directory.

Extending the Project

Adding New Agents

1. Define the agent in src/crypto_trader/config/agents.yaml
2. Implement the agent class in src/crypto_trader/crew.py
3. Add associated tasks in src/crypto_trader/config/tasks.yaml

Adding New Tools

1. Create a new tool in src/crypto_trader/tools/
2. Implement the tool following the crewAI BaseTool pattern
3. Add the tool to the appropriate agent in src/crypto_trader/crew.py

Troubleshooting

Common issues:

1. API Key Errors: Ensure all required API keys are properly set in the .env file
2. Dependency Issues: Make sure UV is installed and dependencies are properly locked
3. Rate Limiting: If hitting RapidAPI rate limits, consider implementing retry logic or upgrading your plan

Dependencies

• crewai: Framework for multi-agent AI systems
• requests: For API interactions
• pydantic: For data validation
• python-dotenv: For environment variable management

License

See LICENSE file for details.