Option pricing using data-driven machine learning approaches: empirical evidence from Indian financial market

Option pricing has always been seen as a black box, as it is highly complex, fierce, dynamic, and volatile. This study explored the application of machine learning (ML) models like stochastic gradient descent (SGD), decision tree, random forest, XGBoost, and artificial neural networks (ANN) to forecast the option price of NIFTY 50 index in Indian financial market. And the study determines the best model in comparison to the Black-Scholes Merton (BSM) model in forecasting the option prices. The study has applied a methodology called multi-model option-pricing neutral intelligence grid-search cross-validation assessment (MONICA) to value index options. The findings can help the traders and investors with their buy/sell strategies by comparing the forecasted value with the quoted value. The consistent and robust results from subsets using the same inputs and significance of Diebold and Mariano (DM) test statistics validates the outperformance of random forest and ANN approaches.