Optimizing Portfolio Hedging Strategies for F30 and SPY

In this paper, we propose three optimization methods for portfolio hedging strategies for two assets, F30 and SPY, and evaluate their effectiveness by maximizing the Sharpe ratio. The first approach, based on six different SPY weighting strategies combined with rolling window optimization, identifies the optimal portfolio that achieves the highest risk-adjusted return (Sharpe ratio of 1.5090) under minimized risk conditions. The second approach utilizes Recursive Feature Elimination (RFE) and Gradient Boosted Regression (GBR) models for feature selection and forecasting by introducing richer macro-market data features, which may lead to overfitting in high volatility environments (Sharpe Ratio of 1.3156), although it displays significant advantages in capturing non-linear changes in the market. The third approach directly optimizes the objective function by expanding the rolling window to smooth short-term market volatility and reduce trading frequency and cost, resulting in a final risk-adjusted return of 1.210. The results of the study show that these approaches exhibit better portfolio optimization capabilities under different market conditions, providing new ideas for achieving more robust risk management and return optimization.