2026-09-08 JST. The original45 v2 outcomes have already been inspected. Original defaults are NOT changed or overwritten. User's phrase 'some distance beyond the pivot' and screenshot do not uniquely determine a0.5ATR stop. Consequently it is not valid to conclude the user's discretionary method fails from that buffer alone.

Freeze this supplementary set before running it: stop buffer1.0/2.0/3.0 execution ATR, modes BREAK_NEAR/HYBRID, execution XYZ5m/15m, CASH_RTH/XYZ_UTC pivot sources (24 cases); additionally DP-only LONG XYZ5m CASH_RTH, both modes and all3 widths (6 cases). Total30 exploratory cases. All other original parameters unchanged, including next-pivot TP, entry timing, 6h holding, costs and conservative fills. No best stop will be promoted as validated from this sample. Report all30, early/late stability, and effect of loss/profit weighting. This is an explicit sensitivity extension motivated by an underspecified numerical rule, not evidence of fresh predictive validation.