Forecasting the Stock Index of SBI Using Wave-Nets

Abstract

This thesis presents a new network to forecast the daily stock value of SBI using the technique of the adaptive wavelet-neural-network. This has been implemented by writing programs in MATLAB. The learning algorithm of this network is optimal on rate of convergence and allows tuning the synaptic weights, dilations and translations parameters of wavelet activation functions. Different types of wavelets can be used as the activation function. The simulation of developed wavelet-neural network architecture and its online learning algorithm justifies the effectiveness of the approach. The algorithm which has been implemented in MATLAB is used by varying different parameters. Thus by using different parameters we have generated a network which is very effective in forecasting of the stock value of the SBI.