A new option pricing model using GARCH with filtered historical data was developed. It can better fit market option prices by considering different return dynamics. The model outperformed other models in pricing accuracy, thanks to its flexible measure change, asymmetric volatility, and nonparametric innovation distribution. The model also aligns with economic theory by showing decreasing state-price densities. Implied volatility smiles are explained by the model's asymmetric volatility and negative skewness of historical data.