Short term stock prediction using SOM

Abstract

In this paper, we propose a stock movement prediction model using self organization map. The correlation is adapted to select inputs from technical indexes. The self-organization map is utilized to make decision of stock selling or buying. The proposed model is tested on the Microsoft and General Electric. Through the experimental test, the method has correctly predicted the movement of stock with close to 90% accuracy in trainnig dataset and 75% accuracy in datatest. The results can be further improved for higher accuracy. © 2009 Springer Berlin Heidelberg.