Estimating the Hurst Exponent with Rescaled Range Analysis
Summary
The Hurst exponent is presented as a measure of long-term dependence in a time series. Values above 0.5 are associated with persistence and possible trending behavior, values below 0.5 with anti-persistence, and a value near 0.5 with random-walk behavior. These interpretations can help researchers form hypotheses about market dynamics, but they do not by themselves establish a profitable trading signal.
The article demonstrates rescaled-range estimation: split observations into chunks of different sizes, center each chunk around its mean, cumulatively sum the deviations, measure the range, and divide that range by the standard deviation. Averaging the ratios by chunk size and fitting a line to log range ratios against log sizes gives an estimated exponent as the slope. An eight-observation illustration produces an estimate of 0.87, which the article interprets as persistent, while explicitly cautioning that the sample is far too small for a sound conclusion. Practical estimates require substantially longer data and can be sensitive to sample length and method choices.
Key ideas
- The Hurst exponent summarizes long-term memory in a time series.
- Values above or below 0.5 are commonly interpreted as persistence or anti-persistence, respectively.
- Rescaled-range analysis compares cumulative deviations with their standard deviation across chunk sizes.
- The exponent is estimated as the slope relating log chunk size to log average rescaled range.
- Small samples can produce misleading estimates, so the example is illustrative rather than conclusive.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.