Algorithmic Trading Strategy using Backtesting & Risk Analysis

Welcome to my project repository for developing and evaluating an algorithmic trading strategy using historical financial and macroeconomic data. The goal of this project was to create a robust trading model that integrates technical indicators and macroeconomic signals, and then test its performance through backtesting and risk evaluation.

• Introduction
• Project Objectives
• Data Acquisition and Preparation
• Feature Engineering
• Modeling Approach
• Backtesting and Risk Analysis
• Results and Insights
• Conclusion
• How to Run the Project

In this project, I developed an algorithmic trading strategy focused on the QQQ ETF (tracking the NASDAQ-100). The strategy incorporates both technical and macroeconomic indicators and uses machine learning to generate buy/sell signals. I evaluated the performance through rigorous backtesting and drawdown analysis.

• Retrieve and clean historical financial and macroeconomic data
• Engineer relevant trading features using TA-Lib and custom formulas
• Train a predictive model using XGBoost for market movement classification
• Backtest the strategy over historical periods and benchmark against a buy-and-hold strategy
• Analyze performance, returns, and risk metrics

• Sources:

  • Yahoo Finance (yfinance) for QQQ historical data
  • FRED API for macroeconomic indicators (VIX, CPI, Fed Rate, etc.)

• Cleaning:

  • Forward-filled missing macroeconomic data
  • Aligned indicators with price time series

• Technical Indicators: RSI, MACD, Bollinger Bands, SMA/EMA
• Macroeconomic Features: VIX Index, Industrial Production, Fed Rate, CPI
• Rolling Features: Lagged returns, volatility, and trend strength metrics

• Model: XGBoost Classifier to predict binary labels (buy or no-buy signal)
• Evaluation:
  • Train/test split with TimeSeriesSplit
  • Metrics: Accuracy, Precision, Confusion Matrix

• Applied trading logic based on predicted signals
• Calculated daily returns from QQQ price data
• Assessed cumulative returns and drawdowns
• Compared results against a buy-and-hold benchmark
• Plotted performance curves and risk metrics

• The strategy captured major uptrends and avoided significant drawdowns
• XGBoost improved predictive accuracy when macro indicators were added
• Cumulative returns outperformed passive investing in several market phases
• Strategy is robust but sensitive to feature scaling and economic shocks

This project demonstrates the effectiveness of combining technical and macroeconomic signals with machine learning for algorithmic trading. While the strategy shows promise, further enhancements could include real-time signal processing and additional risk controls.

1. Clone the repository
2. Install required libraries:

   pip install -r requirements.txt

3. Open Algorithmic Trading Strategy using Backtesting & Risk Analysis.ipynb
4. Insert your FRED API key if prompted
5. Run all cells sequentially

I hope you find this project insightful and valuable for your algorithmic trading strategies. Contributions and feedback are welcome!