Stock Price Prediction Based on Hierarchical Structure of Financial Networks

详细信息    查看全文

• 作者：Kanghee Park (20)
  Hyunjung Shin (20)
  
• 关键词：Stock Price Forecasting ; Hierarchical Structure ; Semi ; Supervised Learning (SSL) ; Machine Learning
• 刊名：Lecture Notes in Computer Science
• 出版年：2013
• 出版时间：2013
• 年：2013
• 卷：8227
• 期：1
• 页码：465-472
• 全文大小：441KB
• 作者单位：Kanghee Park (20)
  Hyunjung Shin (20)
  
  20. Department of Industrial Engineering, Ajou University, Wonchun-dong, Yeongtong-gu, Suwon, 443-749, South Korea
  
• ISSN：1611-3349

文摘

Stock prices are influenced by many external factors such as the oil prices, the exchange rates, the money interest rates, the certificate of deposit (CD), the gold prices, the exchange rates, the composite indexes in global markets, and so on. And the influence among these factors is reciprocal, cyclic, and often hierarchical, which can be naturally presented as a network. In this paper, a prediction method based on hierarchical structure of financial networks is proposed. Semi-supervised learning (SSL) is employed as a base algorithm, and revised to be suited for time series prediction. A network consists of nodes of the factors and edges of similarities between them. The layered structure of networks is implemented by reforming the existing integration method for multiple graphs. With the hierarchical structure of financial networks, it is able to reflect the complicated influences among the factors to prediction. The proposed method is applied to the stock price prediction from January 2007 to August 2008, using 16 global economic indexes and 200 individual companies listed to KOSPI200.