Artificial Neural Network, Back Propagate Algorithm, Stock Market Index
The project studies the predicting ability of an artificial neural network employing the back propagate training algorithm. In the literature, the back propagate neural network has been widely applied to financial industry for solving different problems. This project focuses on the prediction of Shanghai Composite Index movement using the BPN algorithm. The experiment results show that the ANN with BP algorithm is able to capture the index future movement, although there are high tracking errors in the both in-sample and out-of-sample period. The BP algorithm may be replaced by heuristic algorithm for a more precise output.
Over the past few decades, artificial neural networks (ANN) has been widely applied in different industries, for examples, the applications of ANN in medicine, business, economics and finance. Especially when the financial industry becomes more and more dependent on advanced computer science technologies in order to maintain competitiveness in a global economic environment, neural network starts to represent an exciting and reliable technology in many financial applications, e.g., financial conditions analysis, business failure prediction, debt risk assessment, security market applications and financial forecasting. The successful applications of ANN are because the neural network has been motivated right from its inception by the recognition that human brain computers in an entirely different way from traditional digital computer and traditional mathematics computation. Human brain is a highly complex, nonlinear, and parallel information-processing system. The brain has the capability to organize its structural constituents, known as neurons which perform certain computations many times faster than digital computer in existence today. Using this important feature of neurons, researchers construct a paradigm which is composed of a large number of highly interconnected processing neurones network in order to solve some complex problems. In this project we employ ANN's with the BPN algorithm to predict future stock index movement.
This project applies the neural network with back propagation algorithm to study the prediction problem of stock market index movement. Although the simple three-layer neural network cannot track the in-sample market index prices precisely, it is able to capture the out-of-sample market index movement. To enhance the output precision of the neural network, applying heuristic alogrithms instead of BP is considered in the future work.