Hurstify
new Hurstify(config?)Create an estimator with scaleA1, scaleA2, scales, sampleSize, iterations, blockSize, optimizerType, hMin, and hMax.
estimateSingle(window, opts?)Return the estimated Hurst parameter for one numeric series.
estimateSingleWithDiagnostics(window, opts?)Return H, minimizedD, and diagnostic values for one series.
rolling(series, windowSize, step?, onProgress?)Estimate H across rolling windows.
estimateBatch(windows)Estimate a collection of windows and return per-window errors safely.
Synthetic generators
generateFractionalBrownianMotion(n, h)Generate a fractional Brownian motion path.
generateFractionalNoise(n, h)Generate fractional Gaussian noise.
generateGaussianBatch(n)Generate a batch of standard Gaussian values.
generateCorrelatedGaussian(n, rho)Generate correlated Gaussian values for simulations.
Inference
getAsymptoticVariance(H, n)Estimate asymptotic variance for the Hurst estimator.
getStandardError(H, n)Return the standard error implied by the asymptotic variance.
getConfidenceInterval(H, se, level?)Construct a confidence interval around H.
ksPvalue(D, n1, n2)Compute a two-sample KS p-value.
runKalmanFilter(observations, options?)Smooth a noisy H trajectory.
Data and preprocessing
parseCsv(csv)Parse CSV input into numeric records.
extractSeries(rows, key?)Extract a numeric time series from records.
computeRealizedVariance(prices, interval?)Compute realized variance from price observations.
applyPreprocessingPipeline(series, steps)Compose supported centering, standardization, transform, and debiasing steps.
Optimizers, models, and utilities
runBrent / runNelderMead / runAdaptiveGridSearchRun a supported optimizer directly.
RoughBergomiModel / RoughFsvModel / FractionalOuModelSimulate rough-volatility model paths.
setRandomSeed(seed) / resetRandomSeed()Make supported generation and estimation workflows reproducible.
computeKsDistance(a, b)Compute the Kolmogorov–Smirnov distance between samples.
This reference is maintained as product documentation. TypeScript declaration files and source remain canonical for exact overloads and advanced generic types.