R/S analysis and DFA: finite sample properties and confidence intervals

摘要: We focus on finite sample properties of two mostly used methods Hurst exponent H estimation – R/S analysis and DFA. Even though both have been widely applied different types financial assets, only several papers dealt with which are crucial as the differ significantly from asymptotic ones. Recently, has shown to overestimate when compared However, we show random time series lengths 2^9 2^17 that even estimates truly higher than an limit 0.5, they remain very close proposed by Anis & Lloyd estimated standard deviations lower ones On other hand, DFA 0.5. The results propose still remains useful robust method newer is usually preferred in recent literature.