Performance evaluation of Neural Network approach in financial prediction: Evidence from Indian Market

摘要

This paper performs an experiment to forecast stock market movement in India using Artificial Neural Network (ANN) tools and a comparative study is made for selected stock indices to find the optimal selection of parameters in ANN approach. Our objective is to apply the ANN applications in financial market prediction to check whether ANN models add value and may be worthwhile to undertake a research study in this area. It seems that ANN models can indeed offer a potentially rewarding alternative approach. If we design ANN with optimal selection of its parameters (number of neurons in each layer, number of hidden layers, learning rate) then the given result is much better than satisfactory but otherwise its performance and potentiality may not be judged properly.