Associative Classification Model for Forecasting Stock Market Trends

This paper proposes an associative classification model based on three technical indicators to forecast future trends of stock market. Our methodology assessed the performance of nine technical indicators, using a portfolio of ten stocks and a 12-year time series. The experimental results showed that the use of a set of technical indicators leads to higher classification rates compared to the use of sole technical indicators, reaching an accuracy of 88.77%. The proposed approach also uses a multidimensional data cube that allows automatic updating of stock market asset values, which are essential to keep the forecast updated. The results indicate that our approach can support investors and analysts to operate in the stock market.