This paper applies an Adaptive Neuro-Fuzzy Inference System (ANFIS) for improving the estimation of option market prices. The research designs seven ANFIS models, one for each underlying stock, based on the transaction data of the Indian Stock Option with three volatility measures, namely historical volatility, implied volatility, and GARCH volatility. The pricing capability of each model has been compared with the performance of the pure Artificial Neural Net (ANN) and Black-Scholes (BS) model.

Empirical results show that out-of-sample pricing performance of ANFIS is superior to BS, and is also better than pure ANN. The results confirm that the ANFIS model could significantly reduce the Root Mean Square Errors (RMSE) of forecasting, and also provide an alternative way to refine the options' valuation. Further, ANFIS is an interesting alternative for neural modeling which has the same capabilities as standard back-propagation network but is explicit about its decision rules.

Adaptive Neuro-Fuzzy Inference System, ANFIS, Black-Scholes, BS, Artificial Neural Net, ANN, Root Mean Square Errors , RMSE, Bajaj Auto, Hindustan Unilever LimitedHUL, Infosys, Ranbaxy, state bank of india, SBI, Tisco.