CMStatistics 2018: Start Registration
View Submission - CMStatistics
Title: Ordinal pattern dependence in contrast to other concepts of dependence Authors:  Alexander Schnurr - University Siegen (Germany)
Ines Muenker - University of Siegen (Germany) [presenting]
Abstract: Recently, the concept of ordinal pattern dependence was introduced. It is an intuitive measure for the degree of co-monotonic behavior between two time series. The question has been raised what the connection between this concept and other dependence measures (like correlation, Kendall's tau and Spearman's rho) is. We give a partial answer to this question by an empirical study of various types of data sets. The key result is that the new method is suitable to measure dependence and it is much more robust against shocks and/or structural breaks than other dependence concepts. Finally, we shed some light on the applications of ordinal pattern dependence in the context of extremal events.