Status: pre-registered 2026-09-06 (commit af55f80) and run the same day; verdict: FAIL (EURUSD PF 1.11, expectancy +0.050 R; GBPUSD net −0.6 R; see research/2026-09-06-entry5m-v6-usd-scaleout-screen.md); the confirmation run was not triggered. Parents: v3 2026-09-06-entry5m-v3-fixed3r-prereg.md (FAIL 1/3: EURUSD 415 trades net +53.9 R PF 1.21 passed alone; GBPUSD 388 trades net +19.6 R PF 1.08), v5 2026-09-06-entry5m-v5-xau-scaleout-prereg.md (the scale-out mechanics, FAIL on XAUUSD).
"Try v3 divergence for EURUSD and GBPUSD with half profit at 1x and other half at 3x."
| Question | Decision |
|---|---|
| Base rule | v3 exactly: bias_source = "divergence" (ZTD H1 fire, bias until zero touch / contradiction / timeout), stop_mode = "h1_extreme", target_mode = "fixed_r", target_r = 3.0 |
| Scale-out | As v5: half (scale_out_frac = 0.5) exits at entry ± 1.0 x risk (limit, no slippage); the stop on the remaining half moves to the fill price from the next bar (breakeven_after_scale = True, assumed from v5, not re-asked); the rest exits at 3R, at the stop, or at the bias end |
| Instruments | EURUSD and GBPUSD, five-year Tickstory 5m data, 2 ticks per side at 0.00001 |
| Code | None: the run uses flags that exist at HEAD 24aa2de |
pivot_n = 3, min_pb_bars = 3, timeout_h1 = 96, min_rr = 1.0, pending_max_bars = 48, max_hold_bars = 0, stop_mode = "h1_extreme", target_mode = "fixed_r", target_r = 3.0, bias_source = "divergence", scale_out_r = 1.0, scale_out_frac = 0.5, breakeven_after_scale = True, ZTD defaults, on EURUSD and GBPUSD.
EURUSD and GBPUSD share a dollar driver and are one test (v1 spec section 3). Because only that one group is run, the judged rule is stricter than the group rule: both symbols must individually clear all five section-8 sub-criteria (at least 200 trades; PF > 1.15 and ex-outlier PF > 1.0; expectancy > +0.05 R; best month <= 35% of net R; both calendar halves net-positive). The per-symbol verdict column is the judged line; the "overall: n of 3 groups" line does not apply.
A pass is provisional: the pre-declared follow-up is the same cell on XAUUSD and EURJPY, judged with the usd group by the two-of-three group rule, before the rule is called alive. A fail means the variant is dead as specified; no parameter search follows. The sensitivity grid is printed and NOT judged.
Expected side effects, stated now (from v5): win rate up, average win in R down, fewer full-size losses, more trades as breakeven exits free the position earlier. On v3's EURUSD numbers the direction of net R is not predictable in advance; that is the point of running it.