RK-SAVR estimator
Multi-scale Kolmogorov–Smirnov estimation for roughness.
Zero-dependency JavaScript
Measure the roughness of volatility.
Estimate the Hurst parameter directly in Node.js or the browser with a production-ready implementation of RK-SAVR.
✓ Zero runtime dependencies✓ ESM · CJS · IIFE✓ Browser + Node.js✓ TypeScript declarations
A single parameter0 → 1
0.10rough
0.50Brownian
0.80persistent
Lower H means rougher, more anti-persistent motion. Higher H means smoother, more persistent motion.
Interactive playground
Try the estimator now.
A real browser call to Hurstify. Generate a known path or paste your own time series.
Presets
Advanced settings
Result
Run an estimate to see diagnostics.
Your generated path will appear here.
KS distance · —
What does H mean?
One number, three useful intuitions.
H ≈ 0.1
rough / anti-persistent
H ≈ 0.5
Brownian-like
H ≈ 0.8
smooth / persistent
Minimal example
Estimate H in one call.
The demo uses the same public API you can import in your own app. The package is published from this repository with no runtime dependencies.
Open the full guideexample.mjs
import { Hurstify, generateFractionalBrownianMotion } from 'hurstify';
const path = generateFractionalBrownianMotion(2000, 0.1);
const result = new Hurstify({
scaleA1: 1, scaleA2: 25, sampleSize: 500, iterations: 16,
}).estimateSingleWithDiagnostics(path);
console.log(result.H);Why hurstify
A focused tool for understanding roughness.
Synthetic generators
Create reproducible fractional Brownian paths for experiments.
Statistical inference
Diagnostics, confidence tools, and significance testing are available.
Browser + Node
Pure JavaScript with ESM, CJS, IIFE, and TypeScript declarations.
Rolling estimation
Track H across windows without leaving your JavaScript workflow.
Zero runtime dependencies
A small, auditable implementation with no production dependency tree.
How it works
Seven steps from increments to inference.
RK-SAVR uses rescaling invariance and randomized Kolmogorov–Smirnov comparison to estimate H without asking the homepage to become a textbook.
- 01
Segmentation
Choose a stationary window.
- 02
Increments
Compare increments at two or more scales.
- 03
Permutation
Reduce serial dependence with block randomization.
- 04
Subsampling
Draw manageable samples at each scale.
- 05
Rescaling
Apply the self-similar H scaling.
- 06
KS minimization
Find H at the smallest distributional distance.
- 07
Variance reduction
Repeat and average for a steadier estimate.
Research provenance
Independent implementation. Clear lineage.
Hurstify implements the RK-SAVR algorithm described by Angelini & Bianchi (2025). It is an independent open-source implementation; this project does not imply endorsement by the paper authors.