MODEL ADOPTION #4470: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 5 — improves win rate from 31.5% to 33.0% — E8 Intelligence Research
- 1Independent Researcher, United Kingdom
Abstract
DECISION: REJECT ELEMENT: No element survives — the +1.4pp win-rate gain is not a testable edge under E8 geometry timing; it is a statistical artifact of removing low-frequency signals that never reached live status in this window. The filter's only concrete effect here is reducing exposure (removing two open trades), not improving realized PnL. EXPECTED IMPACT: Win rate unchanged (4W/2L/2O remains), profit unchanged, but trade frequency drops ~10-15% — net negative for absolute returns. The backtest gain is likely regime-dependent and not tied to E8's rhythm logic (confirmation count is a lagging, non-geometric metric). RISK: Overfitting to a short backtest window; confirmation lag degrades entry prices in fast moves (NFP, breakouts), shrinking average R-multiple; filter removes high-volatility winners that E8's fade polarity specifically targets. TEST: Before any reconsideration, run a 90-day replay-bench on the live engine's historical tick data (not shadow) with the filter
Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
Related
- MODEL ADOPTION #4606: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 6 — improves win rate from 31.6% to 35.8% — E8 Intelligence Research
- MODEL ADOPTION #4600: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 6 — improves win rate from 31.5% to 35.6% — E8 Intelligence Research
- MODEL ADOPTION #4177: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 6 — improves win rate from 32.6% to 38.3% — E8 Intelligence Research
- MODEL ADOPTION #4602: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 6 — improves win rate from 31.6% to 35.7% — E8 Intelligence Research
- MODEL ADOPTION #4626: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 6 — improves win rate from 32.2% to 37.3% — E8 Intelligence Research
BibTeX
@misc{caldin2026model,
title = {{MODEL ADOPTION #4470: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 5 — improves win rate from 31.5% to 33.0% — E8 Intelligence Research}},
author = {Andrew Stewart Caldin},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23526108},
url = {https://zenodo.org/records/23526108},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - MODEL ADOPTION #4470: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 5 — improves win rate from 31.5% to 33.0% — E8 Intelligence Research
AU - Andrew Stewart Caldin
PY - 2026
DA - 2026/10/21
PB - Zenodo
DO - 10.5281/zenodo.23526108
UR - https://zenodo.org/records/23526108
AB - DECISION: REJECT ELEMENT: No element survives — the +1.4pp win-rate gain is not a testable edge under E8 geometry timing; it is a statistical artifact of removing low-frequency signals that never reached live status in this window. The filter's only concrete effect here is reducing exposure (removing two open trades), not improving realized PnL. EXPECTED IMPACT: Win rate unchanged (4W/2L/2O remains), profit unchanged, but trade frequency drops ~10-15% — net negative for absolute returns. The backtest gain is likely regime-dependent and not tied to E8's rhythm logic (confirmation count is a lagging, non-geometric metric). RISK: Overfitting to a short backtest window; confirmation lag degrades entry prices in fast moves (NFP, breakouts), shrinking average R-multiple; filter removes high-volatility winners that E8's fade polarity specifically targets. TEST: Before any reconsideration, run a 90-day replay-bench on the live engine's historical tick data (not shadow) with the filter Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
ER -