Reproducibility package — Closed-form universalities for fnancial multifractal cascades: lag-dependent intermittency and the cross-domain Hausdor-dimension bridge

Description (paste into description box):

This Zenodo record is the reproducibility package for the Physica D manuscript "Closed-form universalities for financial multifractal cascades: lag-dependent intermittency and the cross-domain Hausdorff-dimension bridge" by R. A. Satnoianu (2026).

Headline. The manuscript establishes three closed-form universality results for financial multifractal cascades, each parameter-free and confirmed at <1% precision on a 39-instrument basket of 38.6 million minute bars across five asset classes (metals, equities, FX, energy, crypto). First, the Bacry–Muzy intermittency parameter takes the closed form λ²* = ln²φ/(4 ln 2) ≈ 0.0836 at class-specific cascade-attractor scales τ_class. Second, the class-mean τ_class is predictable from three public market-mechanics inputs (dominant break period, dominant cycle period, continuous-trading fraction) at R² = 0.97. Third, the box-counting Hausdorff dimension of the cascade-event set on high-resolution financial data converges to α* = log₂φ ≈ 0.6942 within 0.03% on the basket median at the canonical 2σ threshold — the same closed-form constant that bounds the multifractal-spectrum width in three-dimensional Navier–Stokes turbulence.

Cross-domain α universality.* The same closed-form constant is empirically confirmed across four independent physical systems: 3D Navier–Stokes turbulence (1.3% on JHTDB DNS), financial markets (0.03% on M1 basket), solar-wind plasma (5.8% on WIND/MFI 22 ms data), and prime-gap fluctuations (0.0% at the appropriate threshold on the first 5×10⁶ primes). The four-system universality supports a unifying interpretation: financial markets behave as numerical fluids and prime gaps as arithmetical fluids, both realisations of a common multifractal-cascade process whose closed-form invariants are independent of the underlying support.

Package contents. The 58-file package includes: (i) the manuscript source (LaTeX elsarticle 5p, PDF, markdown), (ii) all five figures in PDF and PNG, (iii) 19 Python analysis scripts covering the n=39 basket builder, per-class λ²(τ) profile estimation, T_class fragmentation regression, box-counting d_H universality with bootstrap and threshold-sensitivity tests, cross-domain extensions to solar wind and prime gaps, leverage regression, and figure generators, (iv) 14 result JSONs containing all per-instrument and per-class statistics, (v) cover letter, research highlights, author declaration, and suggested-reviewer list for the Physica D submission. README.md provides step-by-step reproduction instructions and a numerical-result map linking each manuscript element to its source code and JSON.

Dependencies. Python ≥ 3.9, numpy, pandas, scipy, matplotlib, dukascopy-python, pyarrow, cdflib (for the solar-wind CDF reader).

Companion papers. Lag-dependent intermittency in financial multifractal cascades (PRL letter, reproducibility at 10.5281/zenodo.20005017); Parameter-free upper bound on the multifractal width of three-dimensional turbulence (PRL letter); A golden-ratio grid test for prime-counting residual scaling (in preparation for Journal of Number Theory).