When two opposite strategy.entry calls fire on the same bar from flat —
a market entry first, a priced (stop) entry second — the priced entry's
fill previously took the close-only-opposite gate: it closed the
market entry's position and was itself dropped, leaving the engine flat
where TradingView holds the second entry's position. A clean-room probe
(committed with its TV export) pins TV running no arbitration: both
entries execute, and the second call's size covers its own quantity
plus the pending same-bar opposite market quantity — which the ordinary
full-reversal path reproduces exactly, since the closed position IS the
market entry's quantity when placed from flat.

Fix: a placement-time pending-order flag (priced entry created flat
with an earlier same-bar opposite MARKET sibling) scopes the
close-only-opposite gate out of market-first cells. Pending stop
siblings do not set the flag (stop-vs-stop single-close preserved);
placement-rejected entries never reach the queue; deferred-flip carries
never match (created with a position open), so the gate's original
semantics survive untouched.

The probe's engine trade count converges to exactly TV's 7,429; the
strategy that motivated the class is byte-identical (its dual bars ride
the placement-rejection path). Out of pinned scope, characterization-
frozen: stop-first cells that never ran in the probe, pyramiding>0, and
TV's buy-first intra-open ordering (net position parity reproduced;
trade decomposition of one cell differs, documented).

Corpus 240/11/1 zero movement; full private sweep 0 up / 0 down.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>