Clustering analysis of Dow Jones 30 based on extreme points

Abstract

In this paper, Dow Jones 30 (DJ30) are clustered by emerging clustering method based on the differences of stocks' synchronic extreme points' emerging time and their implied range. This method can be applied to classify numerous and disordered data. During the clustering processes, Entropy Method is used to establish stock-distance by principal component. By linear programming method, we clustered DJ30, the results show that this method can cluster stocks with similar trend together: within clusters the curves of stocks are homogeneous and among clusters the curves of stocks are inhomogeneous.