Improving Accuracy of Stock Indices Predictions using Neural Network Forecasts with Sectorial Focuses

摘要

Currently, there is much room for exploration in the area of sectorial stock market predictions. Different sectors of the economy are subjected to influence of many varied economic, political and financial factors. Each influencing factor can be modelled with different degree of readiness. The aggregate effects of factors acting on one segment of the economy can be very different from the effects on another segment, thus causing stock indexes of various segments to move in different directions and magnitudes. Thus, a better way for stock market analysis is to focus on the different sectors of the stock market rather than predicting the overall market index. In this study, neural network prediction is applied over the different sectors of the Singapore Stock Exchange - such as commerce, finance, property and so on. Input factors for neural consideration vary from one sector analysis to another. Results show that sectorial market prediction is indeed more accurate for certain market segments of the Singapore Stock Exchange.