This project implements a multi-phase Statistical Arbitrage (StatArb) strategy using machine learning, econometrics, and quantitative finance techniques. The strategy identifies and trades pairs of cointegrated assets, leveraging mean-reversion principles and risk management techniques to optimize portfolio performance.

1. Linked Asset Identification
   Utilized K-Means Clustering on key financial features such as returns, volatility, beta, and rolling correlations to identify linked asset pairs.

2. Cointegration Testing
   Applied the Engle-Granger test to determine cointegration between asset pairs, ensuring long-term stability of their price relationships.

3. Pair Trading Strategy
   Constructed a pair trading strategy based on price ratios. Positions (long/short) are opened when the ratio deviates from its historical mean, expecting it to revert over time.

4. Risk Management
   Incorporated Value at Risk (VaR) and Conditional Value at Risk (CVaR) to dynamically manage risk and adjust position sizes based on volatility.

5. Backtesting
   The strategy was backtested using historical data, with performance metrics such as Sharpe Ratio, Maximum Drawdown, and Annualized Returns calculated to evaluate success.