Financial Asset Price Forecasting Based on Intertransaction Association Rules Mining

Abstract

It has been widely accepted that association rules mining, the task of searching for correlations between items in a database, can discover useful rules in stock analysis. Previous studies mainly emphasize on mining intratransaction associations. In this paper, we introduce the concept of intertransaction and the FITI algorithm so that we can effectively forecast the price changes in Chinese capital markets, then we compare FITI with EH-Apriori, and demonstrate the advantages of FITI over EHApriori. At the end of this paper, we apply the algorithm to a dataset of Chinese asset indices and the results indicate the usefulness of intertransaction association rules in price prediction.