LAB #4506 HARMFUL: LEDGER BENCH: Apply filter: NZDUSD SHORT — improves win rate from 31.6% to 33.8% (+2.2pp, n=133) — E8 Intelligence Research
- 1Independent Researcher, United Kingdom
Abstract
IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: NZDUSD SHORT — improves win rate from 31.6% to 33.8% (+2.2pp, n=133)
SAME-WINDOW EFFECT: Applying a blanket NZDUSD SHORT filter would have blocked the one live NZDUSD SHORT (rizzy5m, entry 0.58252) that closed as a WIN, and it would have also blocked the wkjudas NZDUSD SHORT (entry 0.59214) that is still open/REJECTED. This removes a confirmed winner from the window, directly contradicting the backtest's claimed edge. Net effect on this small sample: win rate drops (from 3/6 closed wins to 2/5), and PnL would be negative relative to current state because you are discarding a profitable trade.
OVER-TIME PROJECTION: The +2.2pp improvement is derived from n=133, which is a modest sample but not trivial. However, the live window shows the exact opposite signal — your best NZDUSD SHORT was a winner. Over 3–6 months, a currency-pair-directional filter risks overfitting to a specific regime (e.g., NZD weak
Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
Related
- LAB #4182 HARMFUL: LEDGER BENCH: Apply filter: USDCHF LONG @ LONDON_OPEN with 5 conf — improves win rate from 32.6% to — E8 Intelligence Research
- LAB #4480 HARMFUL: LEDGER BENCH: Apply filter: NZDUSD LONG — improves win rate from 31.5% to 55.6% (+24.0pp, n=117) — E8 Intelligence Research
- LAB #4599 HARMFUL: LEDGER BENCH: Apply filter: Session = NY_OPEN — improves win rate from 31.5% to 38.1% (+6.6pp, n=805 — E8 Intelligence Research
- LAB #4387 HARMFUL: LEDGER BENCH: Apply filter: USDCHF SHORT @ OFF_SESSION with 6 conf — improves win rate from 31.6% to — E8 Intelligence Research
- LAB #4439 HARMFUL: LEDGER BENCH: Apply filter: NZDUSD LONG @ NY_OPEN with 6 conf — improves win rate from 31.5% to 85.7 — E8 Intelligence Research
BibTeX
@misc{caldin20264506,
title = {{LAB #4506 HARMFUL: LEDGER BENCH: Apply filter: NZDUSD SHORT — improves win rate from 31.6% to 33.8% (+2.2pp, n=133) — E8 Intelligence Research}},
author = {Andrew Stewart Caldin},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23553631},
url = {https://zenodo.org/records/23553631},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - LAB #4506 HARMFUL: LEDGER BENCH: Apply filter: NZDUSD SHORT — improves win rate from 31.6% to 33.8% (+2.2pp, n=133) — E8 Intelligence Research
AU - Andrew Stewart Caldin
PY - 2026
DA - 2026/10/22
PB - Zenodo
DO - 10.5281/zenodo.23553631
UR - https://zenodo.org/records/23553631
AB - IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: NZDUSD SHORT — improves win rate from 31.6% to 33.8% (+2.2pp, n=133) SAME-WINDOW EFFECT: Applying a blanket NZDUSD SHORT filter would have blocked the one live NZDUSD SHORT (rizzy5m, entry 0.58252) that closed as a WIN, and it would have also blocked the wkjudas NZDUSD SHORT (entry 0.59214) that is still open/REJECTED. This removes a confirmed winner from the window, directly contradicting the backtest's claimed edge. Net effect on this small sample: win rate drops (from 3/6 closed wins to 2/5), and PnL would be negative relative to current state because you are discarding a profitable trade. OVER-TIME PROJECTION: The +2.2pp improvement is derived from n=133, which is a modest sample but not trivial. However, the live window shows the exact opposite signal — your best NZDUSD SHORT was a winner. Over 3–6 months, a currency-pair-directional filter risks overfitting to a specific regime (e.g., NZD weak Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
ER -