Detecting switching points using asymmetric detrended fluctuation analysis

摘要: Abstract This work uses the concept of Asymmetric Detrended Fluctuation Analysis (A-DFA) to investigate and characterize occurrence trend switching in financial series. A-DFA introduces two new roughness exponents, H + − , which differ from usual one by separately taking into account contributions fluctuations according whether local is, respectively, upward or downward. The developed methodology requires evaluation values ( t ) restricting size largest window around value . We show that behave differently neighborhoods points (SPs) where trends change sign. Properly taken differences between shifted allow identify SP’s. Tests with Weiertrasse functions, isolated peaks, actual series are presented, supporting validity proposed method.