A0830
Title: Ordinal pattern based time series analysis
Authors: Herold Dehling - Ruhr-University Bochum (Germany)
Alexander Schnurr - University Siegen (Germany)
Ines Nuessgen - University of Siegen (Germany)
Jeannette Woerner - TU Dortmund (Germany)
Jannis Buchsteiner - Ruhr-University Bochum (Germany)
Annika Betken - University of Twente (Netherlands) [presenting]
Annika Betken - Ruhr-Universitat Bochum (Germany)
Abstract: In time series analysis, ordinal patterns describe the relative position of consecutive realizations generated by a stochastic process. Among other things, estimators for the probabilities of occurrence of ordinal patterns (ordinal pattern probabilities) in time series are considered. Statistical properties of these estimators in discrete-time Gaussian processes with stationary increments are investigated. By means of Rao-Blackwellization, further the estimation of ordinal pattern probabilities is improved. Moreover, limit theorems that describe the asymptotic distribution of the considered estimators are established. The limit distributions may differ depending on the behaviour of the data-generating processes' autocorrelation function. As an application, the Zero-Crossing estimator is discussed for the Hurst parameter characterizing fractional Brownian motions.
