Comparative Predictive Modeling on CNX Nifty with Artificial Neural Network

Abstract

CNX Nifty being an important barometer to indicate country's growth has always been followed with lots of interest from both academia and industry. Now, CNX Nifty could be predicted or not on a random basis gives rise to many a questions. This sounds redolent with any predictive modeling, though with a certain degree of accuracy inbuilt into the system. The major point of consideration is that predictive modeling could be done by various measures and mechanisms. In predictive modeling Multiple Adaptive Regression (MARS), Classification and Regression Trees (CART), Logistic OLS or Non Linear OLS could be used. Here in this study, the researcher has utilized Neural Network as a "Predictive Modeler" to predict CNX Nifty closing on certain definite time zones under consideration, because it is closer to the functioning of the human brain in comparison to the other models. As Indian markets are a clear case of the weak form of efficiency, so, Neural Network will be an ideal tool for detection or prediction in this market.