README: Hybrid Trading Strategy with LSTM and XGBoost

Overview

This Python script implements a hybrid trading strategy that uses both deep learning (LSTM) and machine learning (XGBoost) models to predict stock price movements and execute trades automatically. The strategy fetches real-time stock data from the Alpaca API and applies advanced technical indicators such as Bollinger Bands, MACD, and On-Balance Volume (OBV) for feature engineering. It executes trades for multiple stock tickers (e.g., AAPL, GOOGL, TSLA), leveraging the strengths of both LSTM for time-series analysis and XGBoost for non-linear relationships.

---

Features

1. Deep Learning (LSTM): Captures temporal dependencies in stock price data, making it suitable for time-series predictions.
2. Machine Learning (XGBoost): Complements LSTM by handling tabular data and non-linear relationships.
3. Hybrid Model: Combines LSTM and XGBoost predictions to enhance accuracy.
4. Technical Indicators: Includes Bollinger Bands, MACD, OBV, and ATR for better market analysis.
5. Real-Time Data: Fetches stock market data in real time using Alpaca API, allowing the script to react dynamically to the latest market conditions.
6. Automatic Trade Execution: Uses limit orders for both long and short positions, minimizing slippage and optimizing trade execution.
7. Periodic Model Retraining: Retrains the models periodically to ensure the strategy adapts to changing market conditions.

---

Requirements

• Python 3.x
• Required Python libraries:
  • alpaca-trade-api
  • pandas
  • numpy
  • tensorflow (for LSTM)
  • xgboost
  • scikit-learn
  • yfinance

You can install the necessary libraries using the following command:

pip install alpaca-trade-api pandas numpy tensorflow xgboost scikit-learn yfinance

---

Setup

1. Alpaca API Setup:

   • Create an account at Alpaca to get your API Key and Secret Key.
   • Insert your API credentials into the script:

   API_KEY = "your_alpaca_api_key"
   SECRET_KEY = "your_alpaca_secret_key"

2. Stock Ticker Configuration:

   • Add the stock tickers you want to trade in the tickers list:

   tickers = ['AAPL', 'GOOGL', 'TSLA']  # Add tickers of your choice

3. Model Retraining Interval:

   • You can adjust how frequently the model retrains itself by modifying the retrain_interval parameter (e.g., every 10 minutes):

   retrain_interval = 60 * 10  # Retrains the model every 10 minutes

---

How to Run

1. Run the Script:

   • Make sure all dependencies are installed, and the Alpaca API keys are configured correctly.
   • Run the script in your terminal or command prompt:

   python hybrid_trading_strategy.py

2. Monitoring:

   • The script will print the portfolio value at regular intervals. You can monitor the progress and stop the script at any time by using Ctrl+C.

---

Key Functions

1. add_technical_indicators:

   • Adds technical indicators like Bollinger Bands, MACD, OBV, and ATR to the raw stock data, enriching the dataset for both the LSTM and XGBoost models.

2. create_lstm_model:

   • Defines the architecture of the LSTM model, including layers and dropout for reducing overfitting.

3. train_hybrid_model:

   • Trains both LSTM and XGBoost models using historical stock data and technical indicators as features.

4. get_latest_data:

   • Fetches real-time stock data from Alpaca and applies the same technical indicators used during model training.

5. run_hybrid_strategy:

   • Combines predictions from the LSTM and XGBoost models and executes buy/sell trades based on the averaged prediction.

6. place_limit_order:

   • Executes buy or sell trades with limit orders, optimizing trade execution to reduce slippage.

---

Considerations

• API Limits: Alpaca API has rate limits. Ensure your script complies with these limits to avoid throttling.
• Market Conditions: This strategy is designed for real-time, high-frequency trading and may behave differently in low-volume or highly volatile markets.
• Backtesting: It is recommended to backtest your strategy with historical data before deploying it in live markets.

---

Conclusion

This Python script provides a robust, hybrid trading solution that combines the predictive strengths of LSTM and XGBoost models. It adapts dynamically to changing market conditions by continuously retraining itself, making it a powerful tool for algorithmic trading in real-time stock markets. By leveraging advanced technical indicators and automating trade execution, it offers an effective way to engage in high-frequency trading strategies.