Comparison of Hurst exponent estimation methods

  • Amjad H. Hamza
  • Munaf Y. Hmood
Keywords: Fractional Brownian motion, Hurst exponent, Short memory, Long memory, Self-similarity, Discrete Wavelet, Long-range dependence, Short-range dependence

Abstract

Through recent years many researchers have developed methods to estimate the self-similarity and long memory parameter that is best known as the Hurst parameter. In this paper, we set a comparison between nine different methods. Most of them use the deviations slope to find an estimate for the Hurst parameter like Rescaled range (R/S), Aggregate Variance (AV), and Absolute moments (AM), and some depend on filtration technique like Discrete Variations (DV), Variance versus level using wavelets (VVL) and Second-order discrete derivative using wavelets (SODDW) were the comparison set by a simulation study to find the most efficient method through MASE. The results of simulation experiments were shown that the performance of the methods is relatively close, except for the SODDW method was the most efficient in MASE.

Paper type Categorize your paper under one of these classifications: General review

Published
2021-06-30
How to Cite
Hamza, A. and Hmood, M. (2021) “Comparison of Hurst exponent estimation methods”, Journal of Economics and Administrative Sciences, 27(128), pp. 167-183. Available at: https://jeasiq.uobaghdad.edu.iq/index.php/JEASIQ/article/view/2162 (Accessed: 28July2021).
Section
Statistical Researches