FinRL Trading Platform v2.0

A modern, modular quantitative trading platform built with Python, featuring machine learning strategies, professional backtesting, and live trading capabilities.

🚀 Key Features

• 🤖 Strategy Framework: Multiple quantitative strategies including ML-based stock selection
• 📈 Risk Management: Comprehensive risk controls and position limits
• 💰 Live Trading: Alpaca integration with paper and live trading support
• 🔧 Modular Design: Clean, extensible architecture following best practices

🏗️ Project Architecture

finrl-trading/
├── src/
│   ├── config/           # Centralized configuration management
│   │   └── settings.py   # Pydantic-based settings with environment variables
│   ├── data/            # Data acquisition and processing
│   │   ├── data_fetcher.py     # Multi-source data integration (Yahoo/FMP/WRDS)
│   │   ├── data_processor.py   # Data cleaning and feature engineering
│   │   └── data_store.py       # SQLite-based data persistence
│   ├── backtest/      # Backtesting system
│   │   └── backtest_engine.py  # Professional backtesting engine powered by bt library
│   ├── strategies/      # Trading strategies
│   │   ├── base_strategy.py    # Abstract strategy framework
│   │   └── ml_strategy.py      # ML-based stock selection
│   ├── trading/         # Live trading execution
│   │   ├── alpaca_manager.py     # Alpaca API integration
│   │   ├── trade_executor.py     # Order execution and risk management
│   │   └── performance_analyzer.py  # Performance analysis
│   └── main.py         # CLI entry point
├── examples/           # Examples and tutorials
│   ├── FinRL_Full_Workflow.ipynb  # Complete workflow tutorial (recommended)
│   └── README.md       # Examples documentation
├── data/               # Runtime data storage (gitignored)
├── logs/               # Application logs (gitignored)
├── requirements.txt    # Python dependencies
└── setup.py           # Package installation

🛠️ Installation & Setup

Prerequisites

• Python 3.11+
• Alpaca Account (for live trading)
• Data Source APIs:
  • FMP API Key

Quick Start

1. Clone the repository

   git clone https://github.com/your-username/FinRL-Trading.git
   cd FinRL-Trading
   

2. Install dependencies

   pip install -r requirements.txt
   

3. Configure environment variables

   # Copy configuration template
   cp .env.example .env
   
   # Edit .env file with your API keys
   # Windows: notepad .env
   # Linux/Mac: nano .env
   

4. Run example tutorial

   # Launch Jupyter Notebook (recommended starting point)
   jupyter notebook examples/FinRL_Full_Workflow.ipynb
   

Complete Example Tutorial

The project includes a comprehensive interactive tutorial covering the entire workflow from data acquisition to live trading:

# View examples documentation
cat examples/README.md

# Run complete workflow tutorial (recommended)
jupyter notebook examples/FinRL_Full_Workflow.ipynb

Tutorial Contents:

• ✅ S&P 500 components data acquisition
• ✅ Fundamental and historical price data fetching
• ✅ Machine learning stock selection strategy implementation
• ✅ Professional backtesting (with VOO/QQQ benchmark comparison)
• ✅ Alpaca Paper Trading execution

📖 Usage Examples

Data Acquisition

from src.data.data_fetcher import get_data_manager

# Initialize data manager (automatically selects best available source)
manager = get_data_manager()

# Check current data source
info = manager.get_source_info()
print(f"Current data source: {info['current_source']}")
print(f"Available sources: {info['available_sources']}")

# Get S&P 500 components
components = manager.get_sp500_components()

# Fetch fundamental data
tickers = ['AAPL', 'MSFT', 'GOOGL']
fundamentals = manager.get_fundamental_data(
    tickers, '2020-01-01', '2023-12-31'
)

# Fetch historical price data
prices = manager.get_price_data(
    tickers, '2020-01-01', '2023-12-31'
)

Strategy Development

from src.strategies.ml_strategy import MLStockSelectorStrategy
from src.strategies.base_strategy import StrategyConfig

# Create ML-based stock selection strategy
config = StrategyConfig(
    name="ML Stock Selector",
    parameters={
        'model_type': 'random_forest',
        'top_n': 30,
        'sector_neutral': True
    },
    risk_limits={'max_weight': 0.1}
)

strategy = MLStockSelectorStrategy(config)

# Generate portfolio weights
data = {
    'fundamentals': fundamentals,
    'prices': prices
}
result = strategy.generate_weights(data)
print(result.weights.head())

Strategy Backtesting

from src.backtest.backtest_engine import BacktestEngine, BacktestConfig

# Configure backtest parameters
backtest_config = BacktestConfig(
    start_date='2020-01-01',
    end_date='2023-12-31',
    initial_capital=1000000,
    rebalance_freq='Q',  # Quarterly rebalancing
    transaction_cost=0.001,  # 0.1% transaction cost
    benchmark_tickers=['VOO', 'QQQ']  # Benchmark comparison
)

# Run backtest
engine = BacktestEngine(backtest_config)
result = engine.run_backtest(
    strategy_name="ML Stock Selector",
    weight_signals=ml_weights,
    price_data=prices
)

# View backtest results
print(f"Total Return: {result.metrics['total_return']:.2%}")
print(f"Annualized Return: {result.annualized_return:.2%}")
print(f"Sharpe Ratio: {result.metrics['sharpe_ratio']:.2f}")
print(f"Max Drawdown: {result.metrics['max_drawdown']:.2%}")

# Generate visualization report
engine.plot_results(result)

Live Trading

from src.trading.alpaca_manager import create_alpaca_account_from_env, AlpacaManager
from src.trading.trade_executor import TradeExecutor, ExecutionConfig

# Connect to Alpaca
account = create_alpaca_account_from_env()
alpaca_manager = AlpacaManager([account])

# Configure execution settings
exec_config = ExecutionConfig(
    max_order_value=100000,
    risk_checks_enabled=True
)

executor = TradeExecutor(alpaca_manager, exec_config)

# Execute portfolio rebalance
target_weights = {'AAPL': 0.3, 'MSFT': 0.3, 'GOOGL': 0.4}
result = executor.execute_portfolio_rebalance(target_weights)

print(f"Orders placed: {len(result.orders_placed)}")
print(f"Execution success: {result.success}")

🎯 Core Components

Data Layer (src/data/)

• Multi-Source Data Manager (data_fetcher.py): Intelligent data source selection and management
  • Yahoo Finance: Free financial data (default)
  • FMP (Financial Modeling Prep): High-quality paid data (requires API Key)
  • WRDS: Academic database (requires credentials)
• Data Processor (data_processor.py): Feature engineering, data cleaning, and quality checks
• Data Storage (data_store.py): SQLite-based data persistence with caching and version control

Strategy Framework (src/strategies/)

• Base Strategy (base_strategy.py): Abstract framework for custom strategies
• ML Strategy (ml_strategy.py): Random Forest-based stock selection

Implemented Strategies:

• Equal Weight Strategy
• Market Cap Weighted Strategy
• ML-based Stock Selection
• Sector Neutral ML Strategy

Backtesting System (src/backtest/)

• Professional Backtesting Engine (backtest_engine.py): Powered by bt library
  • Comprehensive performance and risk analysis
  • Multiple benchmark comparison (SPY, VOO, QQQ, etc.)
  • Transaction cost simulation
  • Visualization report generation

Trading System (src/trading/)

• Alpaca Integration (alpaca_manager.py): Alpaca API client with multi-account support
• Trade Executor (trade_executor.py): Order management and risk controls
• Performance Analyzer (performance_analyzer.py): Real-time position tracking and P&L calculation

Configuration System (src/config/)

• Pydantic Settings (settings.py): Type-safe configuration with environment variables
• Multi-environment Support: Development, testing, production configurations
• Centralized Management: All settings in one place

🔧 Configuration

The platform uses Pydantic-based settings with environment variable support:

Environment Variables

Create a .env file and configure the following variables:

# Application
ENVIRONMENT=development
APP_NAME="FinRL Trading"

# Alpaca API (Required for live trading)
APCA_API_KEY=your_alpaca_key
APCA_API_SECRET=your_alpaca_secret
APCA_BASE_URL=https://paper-api.alpaca.markets  # Paper Trading

# Data Sources (Optional, prioritized: FMP > WRDS > Yahoo)
FMP_API_KEY=your_fmp_api_key           # Financial Modeling Prep


# Risk Management
TRADING_MAX_ORDER_VALUE=100000         # Maximum order value
TRADING_MAX_PORTFOLIO_TURNOVER=0.5     # Maximum portfolio turnover
STRATEGY_MAX_WEIGHT_PER_STOCK=0.1      # Maximum weight per stock

# Data Management
DATA_CACHE_TTL_HOURS=24                # Cache TTL in hours
DATA_MAX_CACHE_SIZE_MB=1000            # Maximum cache size in MB

Configuration Usage

from src.config.settings import get_config

config = get_config()
print(f"Environment: {config.environment}")
print(f"Database: {config.database.url}")
print(f"Risk Limits: {config.trading.max_order_value}")

📊 Performance Metrics

The backtesting engine provides comprehensive quantitative analysis:

Return Metrics

• Total Return: Cumulative portfolio performance
• Annualized Return: Time-weighted annual performance
• Alpha: Excess return over benchmark

Risk Metrics

• Volatility: Standard deviation of returns
• Sharpe Ratio: Risk-adjusted returns (Return ÷ Volatility)
• Sortino Ratio: Downside risk-adjusted returns
• Maximum Drawdown: Peak-to-trough decline
• Calmar Ratio: Return ÷ Maximum Drawdown

Tail Risk Measures

• Skewness & Kurtosis: Return distribution characteristics

Benchmarking

• Information Ratio: Active return ÷ Tracking error
• Beta: Portfolio sensitivity to market
• Tracking Error: Standard deviation of active returns

🤝 Contributing

We welcome contributions! Please follow these guidelines:

Development Workflow

1. Fork the repository
2. Create a feature branch

   git checkout -b feature/your-feature-name
   

3. Install development dependencies

   pip install -r requirements.txt
   pip install pytest black flake8 mypy
   

4. Make your changes with proper testing
5. Commit and push

   git commit -m "Add: your feature description"
   git push origin feature/your-feature-name
   

6. Open a Pull Request

Code Standards

• Type Hints: Use modern Python typing
• Documentation: Add docstrings to all public functions
• Testing: Write tests for new features
• Style: Follow PEP 8 with Black formatting

Adding New Strategies

from src.strategies.base_strategy import BaseStrategy, StrategyConfig, StrategyResult

class MyCustomStrategy(BaseStrategy):
    def generate_weights(self, data, **kwargs) -> StrategyResult:
        # Your strategy logic here
        pass

📋 Roadmap

Completed Features ✅

• ✅ Modular strategy framework
• ✅ ML-based stock selection strategies
• ✅ Professional backtesting system (powered by bt library)
• ✅ Alpaca live trading integration
• ✅ Multi-source data support (Yahoo/FMP/WRDS)
• ✅ Comprehensive risk management system
• ✅ Performance analysis and reporting

Planned Enhancements 🚧

• 🔄 Deep reinforcement learning strategies
• 🔄 Alternative data integration
• 🔄 Multi-asset support (crypto, futures)
• 🔄 Advanced portfolio optimization algorithms
• 🔄 Real-time alerting system
• 🔄 Web visualization interface
• 🔄 Docker containerization

📝 License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

⚠️ Important Disclaimer

⚠️ NOT FINANCIAL ADVICE

This software is for educational and research purposes only.

Always consult with qualified financial professionals before making investment decisions. Past performance does not guarantee future results.

📚 References & Acknowledgments

Academic Papers

• Machine Learning for Stock Recommendation - Machine learning approaches to stock selection
• FinRL: Deep Reinforcement Learning Framework - Deep RL framework for quantitative trading
• Portfolio Allocation with Deep Reinforcement Learning - Portfolio optimization research

Open Source Projects

• FinRL - Deep reinforcement learning framework for quantitative trading
• Alpaca-py - Alpaca trading API
• bt - Flexible backtesting framework for Python

Data Sources

• Yahoo Finance - Free financial data
• Financial Modeling Prep - Professional financial data API
• WRDS (Wharton Research Data Services) - Academic financial database
• Alpaca Markets - Brokerage API and market data

---

Built with ❤️ for the quantitative finance community