Internal working note. The loop-4 accountability pass, run on the miss-screen output of 2026-07-26 (9 movers ≥ +30%/30d with $10M+ dollar volume, all unclaimed; universe 983, data as of the Friday 07-24 close). The week's reader story lives in the start-of-week flagship (2026-07-27-market-brief); this file is the part that is about us, and it stays off the reader surface by design.

The headline finding: the biotech-capital-cycle mid-cap trigger is now THREE autopsies overdue, and this week it cost two names at once — TXG +34% and CERT +38%. The 07-13 autopsy prescribed an expression ladder with a written mid-cap entry trigger so "the next TXG resolves to a documented yes/no." The 07-18 autopsy carried it as still-unwritten and named the earnings checkpoint (DHR/MEDP/TMO printed Jul 21–23). The checkpoint came and went, the trigger was never written, and the lane's two mid-caps ran anyway. This is the exact failure mode two prior autopsies named: not failing to see — both names sit ON the biotech-capital-cycle watchlist — but seeing, writing it up, and leaving no entry condition to fire.

• In our data? Yes — both on biotech-capital-cycle (TXG also on ibb).
• Owning thesis? biotech-capital-cycle, explicitly; its own ratification gate (TMO/DHR/MEDP positive 3m + RSI > 60 together) exists, the mid-cap entry leg still doesn't.
• T−1 trigger? The one prescribed on 07-13 and never written. RSI 66–67 on both names now — the easy entry is gone.
• Fix: the trigger gets WRITTEN this cycle or the lane records an explicit "mid-caps are not our leg" verdict — a third carry without either would make this autopsy section a ritual instead of a control. Note the systemic half was fixed today: the in-zone convert-or-pass pipeline that produced the original TXG silence was found physically broken and repaired (10 conversions executed), so a written trigger now has a working path to a position.

Second finding: VG (Venture Global) +36% is a sector-level conviction miss in the war-premium trade. The desk's escalation coverage caught the crude leg (USO +10%/wk; the retired oil-200 scenario's reopen trigger crossed $130 this week) — but the LNG-export leg ran without a written trigger. VG sits on the escalation watchlist; no perspective owns an LNG-export thesis, and roughly a fifth of global LNG transits the strait whose disruption is the macro story of the month. Same shape as the S cybersec miss (07-18): the lane was flagged, the single-name condition was never written.

• In our data? Yes — escalation watchlist.
• Owning thesis? None claims it; nearest frame is the escalation / war-sensitivity family.
• T−1 trigger? Crude bid + Hormuz LNG-share risk together → LNG exporters; visible when USO crossed $130 with VG at RSI ~50s.
• Fix: either an LNG-export leg gets a written condition under the escalation frame, or the frame records "energy expression = crude instruments only, single-name LNG is not our game." Filed as the open question this autopsy leaves.

The rest of the board

Carried from last week — still open

• Biotech-capital-cycle mid-cap trigger — escalated to this week's headline (above). Third carry is not acceptable.
• Refining-lane entry condition (DK row above) — second carry.
• PL candidate question (from the 07-18 mirror screen) — still open, and PL survives the quality screen again this week (below).

The mirror screen (what got cheap while we watched winners)

The quality-on-sale screen's three survivors this week: PL (−57% from high, RSI 28, +25.9% revenue YoY, $58M FCF per EDGAR — surviving a second consecutive week with still no desk look; the 07-18 candidate question remains unanswered), WDAY (−49%, RSI 50, +13.1% YoY, $2.78B FCF), and new this week PATH (−48.6%, RSI 45, +12.7% YoY, $352M FCF, trend still strong-down — a watch-not-touch profile until the trend base forms). Coverage caveat per the screen itself: only 6 of 242 beaten names had fresh-enough fundamentals to grade; 155 quarantined stale.

Verdict + reasoning

Verdict: 0 new coverage misses, 2 conviction misses (TXG+CERT as one lane failure; VG as the war-premium LNG leg), 6 clean passes / not-our-game, 1 carried lane (DK refining). Every missed name was already in our data — the discovery and watchlist nets keep working. The failure that repeats, now for the third consecutive autopsy, is the same one: a flagged lane whose artifact never carries a written entry condition. The standing rule both prior autopsies proposed is hereby restated as the requirement this file checks next week: a thesis or watch that flags a name is not done until it carries either a written trigger or a documented "not our game." Two specific debts against that rule are now named and dated: the biotech mid-cap trigger and the refining-lane condition.

Sources

• deno task market -- miss-screen / stretch-scan / outlier-scan / trigger-check, run 2026-07-26 on the 2026-07-24 settled close — all four are pure reads over research/market-engine/data/summaries/ (fundamentals legs from the cached EDGAR+Massive reconciliation)
• research/investigations/2026-07-18-miss-autopsy-week-ending-2026-07-17.md
• research/investigations/2026-07-13-miss-autopsy-week-ending-2026-07-10.md
• research/issues/2026-07-25-market-brief.md (the EOW reader story) · research/issues/2026-07-27-market-brief.md (the start-of-week flagship)