Automated Day Trading Agents (ADTA)

ADTA leverages CrewAI and OpenAI's LLM Model for market analysis, signal generation, and risk management—operating strictly during US market hours.

Warning

Disclaimer: ADTA provides simulated trading signals for educational purposes only. Real-time trading behaviour may differ. Not intended for actual investment decisions. Consult a qualified financial advisor.

Features

• Real-time market data collection and validation (WIP; might evolve in the future)
• Technical analysis using RSI, MACD, and Bollinger Bands®
• Risk management and portfolio protection
• Data processing with Polars DataFrames
• Market schedule awareness (9:30 AM - 4:00 PM ET, weekdays only)
• Multi-agent system using CrewAI framework

Prerequisites

• Python 3.8+
• OpenAI API key
• Required Python packages:
  • crewai
  • langchain-openai
  • polars
  • pandas
  • numpy
  • tqdm
  • icecream
  • python-dotenv
  • pydantic
  • requests

Installation

1. Clone the repository:

git clone [email protected]:0xnu/day_trading_agents.git
cd day_trading_agents

2. Install dependencies:

## Prerequisites
python3 -m venv .venv
source .venv/bin/activate
uv pip install -r requirements.txt
python3 -m pip install --upgrade pip

## When you finish
deactivate

3. Set up environment variables:

cp .env.example .env
# Edit .env file with your OpenAI and Alpha Vantage API keys

Project Structure

day_trading_agents/
├── scripts/
│   ├── day_trading_agents.py
│   ├── day_trading_agents_advanced.py
│   └── market_news_agent_terminal.py
├── trading_data/
│   ├── signals/
│   │   ├── csv/    # Polars DataFrame storage
│   │   └── json/   # JSON signal storage
│   └── results/    # Trading session results
├── requirements.txt
└── .env

Debug Mode

Enable debug tracking:

# In scripts/day_trading_agents.py
DEBUG = True  # Enables icecream logging

Debug features include:

• Configuration validation
• Agent initialisation tracking
• Signal processing progress
• File storage operations
• Exception details

Market Hours

Disable and bypass market hours:

# In scripts/day_trading_agents_advanced.py
enforce_market_hours=False

Usage

Run the trading system:

python3 -m scripts.day_trading_agents ## Simple

python3 -m scripts.day_trading_agents_advanced ## Advanced

python3 -m scripts.market_news_agent_terminal ## Market News Terminal

The system will:

1. Verify market hours (US Eastern Time)
2. Process each stock with progress tracking
3. Generate and store trading signals
4. Track all operations with debug logging when enabled

Agents (Simple)

MarketDataAgent

Collects and validates real-time market data using GPT-4 Turbo, ensuring data quality and completeness.

TechnicalAnalysisAgent

Analyses market data using technical indicators (RSI, MACD, Bollinger Bands®) to generate trading signals with confidence scores.

RiskManagementAgent

Evaluates trading signals, manages risk parameters, and provides portfolio protection strategies.

Data Models

• Config: System configuration and parameters
• StockInfo: Basic stock information
• MarketData: Market data validation model
• TechnicalSignal: Technical analysis signals with confidence scores

Agents (Advanced)

MarketDataAgent

Handles real-time market data collection and validation using Alpha Vantage API. The agent:

• Retrieves current and historical price data with rate limiting (5 calls/minute)
• Implements caching to minimise API calls (60-second expiry)
• Validates data quality and completeness
• Processes batch requests for multiple stock symbols
• Stores data in both CSV and JSON formats for analysis

TechnicalAnalysisAgent

Performs technical analysis using multiple indicators:

• RSI (Relative Strength Index)

  • Calculates momentum with 14-period default
  • Generates buy signals below 30
  • Generates sell signals above 70

• MACD (Moving Average Convergence Divergence)

  • Uses 12/26/9 standard periods
  • Identifies trend reversals and momentum
  • Generates signals based on line crossovers

• Bollinger Bands®

  • Implements 20-period moving average with 2 standard deviations
  • Generates buy signals at lower band crosses
  • Generates sell signals at upper band crosses

RiskManagementAgent

Evaluates trading signals and manages portfolio risk:

• Analyses confidence scores for each technical signal
• Monitors signal metadata and technical indicators
• Provides risk assessments based on market conditions
• Generates trade recommendations with risk parameters
• Stores assessment results for future analysis

Data Models

Config: Advanced configuration model with settings:

• API keys for OpenAI and Alpha Vantage (SecretStr for security)
• Model selection and parameters (temperature, max tokens)
• File paths for data and signal storage
• Market hours enforcement flags
• Stock symbol list with descriptions
• Debugging and logging settings

StockInfo: Detailed stock information model containing:

• Trading symbol
• Company name
• Market sector
• Market capitalisation
• Average trading volume

MarketData: Market data validation model with temporal features:

• Trading symbol
• Timestamp with timezone awareness
• OHLC (Open, High, Low, Close) prices
• Trading volume
• Data validation status

TechnicalSignal: Technical analysis signal model:

• Trading symbol and timestamp
• Indicator type (RSI, MACD, Bollinger Bands®)
• Signal value and direction (Buy, Sell, Hold)
• Confidence score (0-1 range)
• Technical metadata dictionary containing:
  • RSI values
  • MACD line and signal line
  • Bollinger Band values (upper, middle, lower)
  • Additional indicator-specific data

SignalStorage: Storage handler for trading signals:

• Structured storage in CSV and JSON formats
• Timestamp-based file organisation
• Historical signal retrieval
• Data persistence and backup
• Signal aggregation and filtering

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

This project is licensed under the BSD 3-Clause License.

Citation

@misc{afoadta2024,
  author       = {Oketunji, A.F.},
  title        = {Automated Day Trading Agents (ADTA)},
  year         = 2024,
  version      = {2.0.1},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.14536356},
  url          = {https://doi.org/10.5281/zenodo.14536356}
}

Copyright

(c) 2024 Finbarrs Oketunji. All Rights Reserved.