Enhancing Stock Prices Forecasting System Outputs Through Genetic Algorithms Refinement of Rules-Lists

Abstract

The intent of stock market was to amass capital in an economy and distribute of same to high-yielding return ventures. Recently, stock markets are considered the foremost meeting point of information such as macroeconomic, national and investor. There are significant mechanisms for measuring future progress in the economy and the markets. Studies have revealed that fuzzy logic control (FLC)-based forecasting models rely on the composition of rules-lists, which are often redundant due to poor mapping of their antecedents and conditions to the consequents. This paper introduced a process of refining the rules-lists with the use of genetic algorithm. A refined rules-list was constructed for FLC rules base after the removal of inherent redundancy. To evaluate the proposed enhanced FLC model, the inputs and output variables were opening, highest and closing prices of Dangote Cement Company Shares, respectively. The outcomes showed that rules-lists of the enhanced FLC were shortened to five (5) rules as against the nine (9) rules in the human expert system. Also, the forecasts of enhanced FLC constructed with refined rules-lists were better than those FLC built with human expert system on the basis of mean square error (MSE) and mean absolute percentage error (MAPE) calculated. In the case of MSE, forecasts improved from 24.898% to 75.102%. Similarly, MAPE forecasts accuracy improved from 32.424% to 67.576% for the enhanced FLC against FLC.