New edge proposed: shannon_entropy_regime_filter — E8 Intelligence Research

{ "name": "shannon_entropy_regime_filter", "source": "Claude Shannon information theory / Shannon entropy applied to returns as regime filter (see work on entropy-based market regime detection, e.g. Nawrocki & Harding 2016; also popularized in quant blogs)", "type": "filter_flag", "rule": "Compute Shannon entropy H of the last 20 log-return bins (10 equal-width bins spanning the rolling 100-bar min/max); only take long entries when H < 2.0 bits, and short entries when H > 2.0 bits (or simply block all trades when H exceeds the 80th percentile of the last 200 bars).", "pseudocode": "r = diff(log(close), 1)\nlo, hi = rolling_min(r,100), rolling_max(r,100)\nbins = linspace(lo, hi, 11)\np = histcounts(r[-20:], bins) / 20\nH = -sum(p * log2(p + 1e-12))\nregime_ok = H < rolling_quantile(H, 200, 0.80)\nsignal = base_signal AND regime_ok", "why_it_works": "Low-entropy return sequences indicate trending/predictable microstructure where directional rules have edge, while high-entropy chop Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com