Exploring Information Categories and Artificial Neural Networks Numerical Algorithms in S&P500 Trend Prediction: A Comparative Study

Exploring Information Categories and Artificial Neural Networks Numerical Algorithms in S&P500 Trend Prediction: A Comparative Study

Salim Lahmiri (ESCA School of Management, Casablanca, Morocco), Mounir Boukadoum (Department of Computer Science, University of Quebec at Montreal, Montreal, Québec, Canada) and Sylvain Chartier (School of Psychology, University of Ottawa, Ottawa, Ontario, Canada)
Copyright: © 2014 |Pages: 19
DOI: 10.4018/IJSDS.2014010105
OnDemand PDF Download:
List Price: $37.50


The purpose of this study is to examine three major issues. First, the authors compare the performance of economic information, technical indicators, historical information, and investor sentiment measures in financial predictions using backpropagation neural networks (BPNN). Granger causality tests are applied to each category of information to select the relevant variables that statistically and significantly affect stock market shifts. Second, the authors investigate the effect of combining all of these four categories of information variables selected by Granger causality test on the prediction accuracy. Third, the effectiveness of different numerical techniques on the accuracy of BPNN is explored. The authors include conjugate gradient algorithms (Fletcher-Reeves update, Polak-Ribiére update, Powell-Beale restart), quasi-Newton (Broyden-Fletcher-Goldfarb-Shanno, BFGS), and the Levenberg-Marquardt (LM) algorithm which is commonly used in the literature. Fourth, the authors compare the performance of the BPNN and support vector machine (SVM) in terms of stock market trend prediction. Their comparative study is applied to S&P500 data to predict its future moves. The out-of-sample forecasting results show that (i) historical values and sentiment measures allow obtaining higher accuracy than economic information and technical indicators, (ii) combining the four categories of information does not help improving the accuracy of the BPNN and SVM, (iii) the LM algorithm is outperformed by Polak-Ribière, Powell-Beale, and Fletcher-Reeves algorithms, and (iv) the BPNN outperforms the SVM except when using sentiment measures as predictive information.
Article Preview

1. Introduction

There is a high interest in forecasting the future trends of stock market indices and stock prices, in a desire to reduce the uncertainty associated with investment decision making (Davalos et al., 2009; Sun, 2010; Joseph & Mazouz, 2010; Hammami & Boujelbene, 2012; Lai & Joseph, 2012; Strang, 2012; Lahmiri, 2013; Lahmiri et al., 2014). However, forecasting stock markets is a challenging task since the dynamics of the market are complex and non-linear, and many factors must be accounted such as business and economic conditions, political events and investor’s expectations. The most important factor in predicting stock markets is the quality of the available information used to model the behaviour of the market. Economic information and technical indicators are the most frequently types of information employed to forecast stock markets. Although economic information is widely adopted by the scholars of modern quantitative finance (Ross, 1976), several academic studies suggest that technical analysis may be effective to extract useful information from past and actual market prices to predict future prices (Pruitt & White, 1988). While academic scholars study theoretically and empirically the relationship between economic factors and stock market using statistical linear models, technical analysts examine the market price time series to identify its regularities by extracting its nonlinear patterns. Academic scholars rely on economic information since they use financial economic theory with strong mathematical foundations to explain the relationship between the economy and the stock market. For instance, classical financial economics is based on normative axioms that underlie expected utility theory, risk aversion, rational expectations, and Bayesian updating to predict market returns. In other words, the decision making process of the investor is rational. However, financial economics models often fail to predict stock market movement with economic information. For instance, researchers found that in many situations the investor deviates from rationality (Hirshleifer, 2001). As a result, behavioural finance has been proposed as an alternative approach to explain and predict stock market behaviour by examining the behaviour of the investor. Indeed, according to Kahneman and Tversky (2000), the investor decision-making process is not rational since his behaviour is influenced by past experiences, beliefs, context, the format of information presentation, and incomplete information. To explain changes in stock prices, researchers in behavioural finance developed an alternative theory based on investor sentiment which is one of the most important psychological aspects. The assumption is that investors are subject to sentiment (DeLong et al., 1990). For instance, the investor psychology affects his decision making process. Thus, the investor sentiment influences stocks returns. Baker and Wurgler (2007) define investor sentiment as “a belief about future cash flows and investment risks that is not justified by the facts at hand”. Using linear statistical models, Baker and Wurgler (2006) show that investor sentiment affects the cross section of stock returns.

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing