MODEL ADOPTION #4479: REJECT — LEDGER BENCH: Apply filter: Quarterly phase = DISTRIBUTION — improves win rate from 31.5% — E8 Intelligence Research
- 1Independent Researcher, United Kingdom
Abstract
DECISION: REJECT ELEMENT: No single rule survives hostile review — the DISTRIBUTION-phase filter is regime-dependent and incompatible with E8's counter-trend FADE timing, which already exploits distribution-phase reversals as part of its edge. EXPECTED IMPACT: Win-rate gain of +4.8pp is likely overfit to late-cycle chop; true effect ±3pp. Filter would cut ~30–40% of trades, including profitable momentum signals during distribution phases in divergent-policy FX pairs (e.g., GBPUSD LONG during BoE/Fed divergence). Net PnL could drop despite higher win rate due to fewer high-R trades. RISK: The filter suppresses exactly the FADE signals that E8's Gods-Clock rhythm identifies as high-probability reversals — distribution phases are where fade edges are strongest. Blocking them degrades the model's core alpha. Also, n=113 is too small; 95% CI spans roughly 28–45% win rate, so the observed improvement is within noise. TEST: Before any reconsideration, run a 90-day shadow replay on the
Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
Related
- LAB #4479 PROMISING: LEDGER BENCH: Apply filter: Quarterly phase = DISTRIBUTION — improves win rate from 31.5% to 36.3% ( — E8 Intelligence Research
- LAB #4612 NEUTRAL: LEDGER BENCH: Apply filter: Quarterly phase = DISTRIBUTION — improves win rate from 32.2% to 36.9% ( — E8 Intelligence Research
- LAB #4441 PROMISING: LEDGER BENCH: Apply filter: Quarterly phase = DISTRIBUTION — improves win rate from 31.5% to 36.3% ( — E8 Intelligence Research
- LAB #4145 PROMISING: LEDGER BENCH: Apply filter: Quarterly phase = DISTRIBUTION — improves win rate from 32.5% to 37.3% ( — E8 Intelligence Research
- LAB #4165 NEUTRAL: LEDGER BENCH: Apply filter: Quarterly phase = DISTRIBUTION — improves win rate from 32.6% to 37.6% ( — E8 Intelligence Research
BibTeX
@misc{caldin2026model,
title = {{MODEL ADOPTION #4479: REJECT — LEDGER BENCH: Apply filter: Quarterly phase = DISTRIBUTION — improves win rate from 31.5% — E8 Intelligence Research}},
author = {Andrew Stewart Caldin},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23526290},
url = {https://zenodo.org/records/23526290},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - MODEL ADOPTION #4479: REJECT — LEDGER BENCH: Apply filter: Quarterly phase = DISTRIBUTION — improves win rate from 31.5% — E8 Intelligence Research
AU - Andrew Stewart Caldin
PY - 2026
DA - 2026/10/21
PB - Zenodo
DO - 10.5281/zenodo.23526290
UR - https://zenodo.org/records/23526290
AB - DECISION: REJECT ELEMENT: No single rule survives hostile review — the DISTRIBUTION-phase filter is regime-dependent and incompatible with E8's counter-trend FADE timing, which already exploits distribution-phase reversals as part of its edge. EXPECTED IMPACT: Win-rate gain of +4.8pp is likely overfit to late-cycle chop; true effect ±3pp. Filter would cut ~30–40% of trades, including profitable momentum signals during distribution phases in divergent-policy FX pairs (e.g., GBPUSD LONG during BoE/Fed divergence). Net PnL could drop despite higher win rate due to fewer high-R trades. RISK: The filter suppresses exactly the FADE signals that E8's Gods-Clock rhythm identifies as high-probability reversals — distribution phases are where fade edges are strongest. Blocking them degrades the model's core alpha. Also, n=113 is too small; 95% CI spans roughly 28–45% win rate, so the observed improvement is within noise. TEST: Before any reconsideration, run a 90-day shadow replay on the Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
ER -