Comparing the performance of FA, DFA and DMA using different synthetic long-range correlated time series

TL;DR

This study compares four estimators of long-range correlations in synthetic time series, finding that CDMA generally outperforms others and remains reliable across different generators and scaling ranges.

Contribution

It provides a comprehensive comparison of LRC estimators using various synthetic generators, highlighting the superior performance of CDMA and DFA.

Findings

CDMA has the best performance among tested estimators.

DFA performs slightly worse but remains reliable.

FA performs the worst in all scenarios.

Abstract

Notwithstanding the significant efforts to develop estimators of long-range correlations (LRC) and to compare their performance, no clear consensus exists on what is the best method and under which conditions. In addition, synthetic tests suggest that the performance of LRC estimators varies when using different generators of LRC time series. Here, we compare the performances of four estimators [Fluctuation Analysis (FA), Detrended Fluctuation Analysis (DFA), Backward Detrending Moving Average (BDMA), and centred Detrending Moving Average (CDMA)]. We use three different generators [Fractional Gaussian Noises, and two ways of generating Fractional Brownian Motions]. We find that CDMA has the best performance and DFA is only slightly worse in some situations, while FA performs the worst. In addition, CDMA and DFA are less sensitive to the scaling range than FA. Hence, CDMA and DFA remain "The Methods of Choice" in determining the Hurst index of time series.