Does scale-dependent dispersion suggest anti-persistence, diffusion, or persistence? Walk the evidence clock one stage at a time. This lab teaches the publication contract; it does not simulate returns, platform parity, or trade performance.
Selected expression
estimate H as slope of log(R/S) versus log(window scale), or another declared estimator
Availability rule
A state at t may use only observations available through t; confirmed swings retain their confirmation timestamp and are never relabeled as known at the pivot bar.
calculated · step 1 of 7
Verified synthetic fixture
This lab is paired with datasets/implementation-fixtures.json: 96 deterministic, oldest-to-newest observations spanning oscillating, rising, and moderating regimes. It is teaching data—not market history or predictive evidence.
Canonical latest output (D08-F07-A09)
{
"value": 0.6970566842387678
}
Executable checks cover this result, flat/zero-volume boundaries, invalid timestamps, and prefix causality in both language entry points.