FRACTAL-LAB

Advanced fractal analysis suite for detecting market memory and deviations from random walk behavior.
Features:

Hurst Exponent proxy via variance scaling with dynamic confidence intervals
DFA-Lite (two-scale Detrended Fluctuation Analysis)
Lo-MacKinlay Variance Ratio test
Volatility memory analysis on |returns|
Quality scoring system (A-F grades)
Regime classification: Persistent / Random Walk / Anti-Persistent

Interpretation:

H > 0.55 → Trending behavior (momentum)
H ≈ 0.50 → Random walk (no predictability)
H < 0.45 → Mean-reverting behavior

Important:

Screening tool only — validate results in Python/R
Does not test for short-range dependence (Lo, 1991)
Recommended sample size: N ≥ 300 bars
NOT a trading system — for research and education only

References: Hurst (1951), Lo & MacKinlay (1988), Peng et al. (1994)