Documentation
Understand the signal before you tune the model.
Explanatory guides for using Hurstify in real workflows. Learn the concepts here; use the API Reference when you need every public option and export.
Start here
Run your first estimate, understand the result, and choose sensible defaults.
Read guideWhat H means
Build intuition for anti-persistence, Brownian-like paths, and persistent motion.
Read guideMethodology
Learn how RK-SAVR turns rescaled increments into a reproducible Hurst estimate.
Read guide01 · Start here
Install from source and estimate H.
Hurstify is currently consumed from this repository rather than an npm release. Build it once, then use the same public API shown in the live demo.
1 · Clone and build
Terminal
git clone https://github.com/sachncs/hurstify.git
cd hurstify
npm install
npm run build2 · Run an estimate
example.mjs
import { Hurstify, generateFractionalBrownianMotion } from 'hurstify';
const path = generateFractionalBrownianMotion(2000, 0.1);
const estimator = new Hurstify({
scaleA1: 1, scaleA2: 25, sampleSize: 500, iterations: 16,
});
const result = estimator.estimateSingleWithDiagnostics(path);
console.log(result.H);02 · Concepts
What does H mean?
The Hurst parameter describes how increments scale and how motion relates to its own past. It is a roughness descriptor, not a promise about future prices.
Anti-persistent
Successive movement tends to reverse more often; paths look rougher.
Brownian-like
The familiar memoryless scaling of a Brownian path.
Persistent
Movement is smoother and local direction persists longer.
03 · Methodology
RK-SAVR, in seven steps.
Hurstify is an independent implementation of the RK-SAVR algorithm described by Angelini & Bianchi (2025). The estimator compares rescaled increment distributions with a randomized two-sample Kolmogorov–Smirnov distance.
- 01
Segment the window
- 02
Compute increments
- 03
Permute blocks
- 04
Subsample increments
- 05
Rescale by H
- 06
Minimize KS distance
- 07
Average iterations