An extension of DFA that characterizes the multifractal scaling behaviour of a time series.Multifractal detrended fluctuation analysis (MF-DFA) is a statistical technique used to examine the scaling properties of a time series—a sequence of measurements taken over time—when those properties vary across different magnitudes or time scales. It extends an earlier method called detrended fluctuation analysis (DFA), which detects long-range correlations in a signal, such as whether a large value tends to be followed by another large value. MF-DFA goes further by quantifying how these correlations change depending on the size of the fluctuations, revealing whether the signal has a 'multifractal' structure, meaning it is composed of many interwoven patterns with different degrees of roughness or variability. This is common in natural phenomena like heartbeats, stock prices, and river flows, where small and large fluctuations behave differently.
MF-DFA was introduced in 2002 by Jan W. Kantelhardt and colleagues, building on the DFA method developed in 1994 by C.-K. Peng and co-workers. It quickly became a standard tool in econophysics and physiology for analyzing nonstationary data.
MF-DFA is used to analyze heart rate variability to distinguish healthy from diseased cardiac function, and to study volatility clustering in financial markets for risk assessment.
MF-DFA has been applied to seemingly unrelated signals, from earthquake seismic records to the rhythm of human gait, and can even detect subtle changes in a person's walking pattern that precede falls in the elderly.











