COMPARISON OF NEURAL NETWORK AND ORDINARY LEAST SQUARES MODELS IN FORECASTING CHINESE STOCK PRICES

Abstract

With the recent entrance of China’s stock market into the global financial system, a renewed interest emerges in discovering the one framework that will reliably forecast changes in stock market prices. This research presents evidence of the forecasting performance of ANN models compared to linear regression models, specifically ordinary least square (OLS) models. Using a 10-year data set from 2002 to 2012 of 154 companies in the A-share of the Shanghai Stock Exchange, the study demonstrates the use of ANNs in forecasting price changes in China’s stock market. The data set used includes observations on daily closing prices and data on 25 indicators, namely, macroeconomic indicators, market sentiment indicators, institutional investors, and microeconomic indicators of factors believed to contribute to the unpredictability of the Chinese stock exchange. A t-test was calculated to compare the performance of the ANN model with the performance of the OLS model in predicting daily stock price change in China’s stock markets.The results of the t-test indicate that the ANN model performed significantly better. Keywords: Business Intelligence, Financial Forecasting, Investment Strategies, Behavioral Finance, Technical Analysis, Data Mining, Neural Networks, and Artificial Intelligence.