The relative performance of option pricing models estimated directly from market data and indirectly from option data has not been subject to many academic studies. In this book the pricing performance of these two estimation methods is tested empirically using options written on the S&P500 Index. A Bayesian approach is taken for the models estimated directly from market data and a regime switching feature is introduced to better capture the market dynamics. This combination results in a new Bayesian pricing algorithm for the regime switching GARCH option pricing model. The resulting model is very well founded from a theoretical perspective, taking into account many frequently observed trends in financial markets. The option inferred models on the other hand, are estimated periodically using implied calibration and represent a practical class of models. These models have been popular among practitioners for a long time due to their straightforward estimation method. This book therefore presents an exciting study that tests a class of theoretical models versus a class of practical models for option pricing performance.