MODEL ADOPTION #4924: REJECT — TG AUTO: New edge proposed: tick_rule_bvp_trend_confirm — E8 Intelligence Research

DECISION: REJECT ELEMENT: Nothing survives as a standalone gate. The only salvageable element is a *diagnostic*, not a filter: log `BVP_20 = sum(sign(close-open)*volume)` and `bar_dir` on every GTX entry for 90 days, but do NOT gate live entries on it. The proposed rule (`bvp>0 AND bar_dir>0` for longs, mirror for shorts) is a trend-confirmation filter bolted onto a mean-reversion/fade engine — that is a category error, not an edge. EXPECTED IMPACT: If adopted as written, expect trade count to drop 20–40% and win-rate to move *unpredictably* — likely flat-to-negative on profit even if win% ticks up, because the filter preferentially kills the counter-trend scalps that are the engine's actual edge (FADE polarity). Merlin's own same-window projection is n≈5 scoreable trades with 3 hypothetical blocks — that is not evidence, it is a story. The 62 confidence is unearned; it should be ~35 given the sample. RISK: (1) BVP on 15s–7m bars is dominated by doji noise — `sign(close-open)` flips Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com