Fitting Multifractal Spectra to Model Trading-Time Cascades

Summary

This article describes a spectrum-fitting stage in a multifractal modeling pipeline. Starting from a scaling function τ(q), it uses a Legendre transform to estimate the singularity spectrum f(α), with a central finite-difference approximation for the derivative. It filters low-q points and implausible or numerically invalid spectrum values, then interprets the curve’s peak and width as indicators of scaling behavior and multifractality.

The empirical spectrum is fitted to four candidate forms associated with normal, binomial, Poisson, and Gamma cascade multiplier distributions. Bounded optimization estimates model parameters, and the model with the lowest sum of squared errors is selected for a later synthetic-path generator. The models differ in symmetry, support, skew, and flexibility. The article is an implementation guide rather than evidence of a profitable trading strategy: fit quality on the observed spectrum does not establish that simulated paths reproduce future market behavior, and numerical filtering and model selection choices can affect results.

Key ideas

• The Legendre transform maps the scaling function τ(q) into a singularity spectrum f(α).
• Central finite differences approximate the derivative when τ(q) is available only at discrete points.
• The implementation filters unstable low-q observations and spectrum values that appear numerically invalid.
• Four parametric spectrum models represent different cascade assumptions and levels of flexibility.
• Bounded optimization selects the candidate with the lowest sum of squared fitting errors for downstream simulation.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.