Hurst Exponent

What is the Hurst Exponent?

The Hurst exponent (H) measures long-range dependence in a series: whether moves tend to continue, partially reverse, or wander at random. H near 0.5 is consistent with a random walk, where past increments say nothing about the next. H above 0.5 indicates persistence, a trending character; H below 0.5 indicates anti-persistence, a mean-reverting character. It is named for Harold Edwin Hurst, who developed the method studying Nile river flows; Mandelbrot later brought it into finance as part of fractal market analysis, where it mirrors fractal dimension (D equals 2 minus H for self-affine series).

The classical estimator is rescaled-range (R/S) analysis: for each window size, divide the range of cumulative deviations from the mean by the standard deviation, then fit the slope of log R/S against log window size; that slope is H. Detrended fluctuation analysis (DFA) is a common alternative that removes local trends before measuring. Both need substantial data, and estimates on trading-sized windows carry wide error bands, so rolling readings are tendencies, not precise numbers.

How traders use it

• As a regime filter: readings persistently above 0.5 support trend-following logic, readings below favor fading and mean reversion, and readings hugging 0.5 argue for standing down both (see Market Efficiency & Regime Persistence Measures for simpler relatives).
• As a rolling persistence gauge: a smoothed H series scales exposure or switches playbooks as the market's character shifts, with smoothing applied because raw estimates jump.
• As a strategy-fit check: estimating H on the instrument and timeframe actually traded tests whether momentum or mean-reversion logic has statistical footing there before any backtest is run.