Detecting Anomalies in the Dynamics of a Market Index with Topological Data Analysis


  • Ngoc Kim Khanh Nguyen Van Lang University
  • Marc Bui 2EA 4004 Human and Artificial Cognition (CHArt) Laboratory, École Pratique des Hautes Études



We investigate the behavior of a market index by using the persistence diagram of its time-delay embedding, a powerful tool of the Topological Data Analysis approach. Furthermore, we propose a framework to capture the characteristics of a market index’s daily return and find its extraordinary movements. Our method bases on the changes in the point distribution of the persistence diagram. After applying for the S&P 500 index in 50 latest years, the framework is demonstrated that it can efficiently track the topological information of the index return’s such that significant anomalies of the time series can be detected.