Miss Autopsy — week ending 2026-07-17
Miss Autopsy — week ending 2026-07-17
Internal working note. The loop-4 accountability pass, run on the miss-screen
output of 2026-07-17 (16 movers ≥ +30%/30d with $10M+ dollar volume; 15 unclaimed,
1 claimed by physical-ai-robotics: CCXI). The week's reader story lives in the
start-of-week flagship brief (2026-07-18-market-brief); this file is the part
that is about us, and it stays off the reader surface by design.
The headline finding: the first genuine COVERAGE miss in three weeks — and it's a theme, not a name. Last week's autopsy found all 19 misses already in our watchlists. This week breaks that streak: the refiner cluster — MPC +29%, DK +53%, CLMT +31% over 30 days — reached our data only through the discovery net (monster-discoveries), not through any thematic energy coverage, and VLO and PSX, the two largest US refiners, are in no watchlist at all. The desk's energy universe covers majors (XOM/CVX), E&P, and services (HAL/SLB/OIH) — but has no refining/crack-spread lane, which is precisely where a war-premium crude tape pays out (product spreads, not barrels). The Friday energy vertical called the re-rate and still couldn't see its hottest expression.
- In our data? Only via the catch-net; zero thematic rows.
- Owning frame? us-energy-dominance (bench) — its "services late-cycle" watch is the adjacent leg; refining is the missing one.
- T−1 trigger? Crude bid + refiners breaking out together (USO up ~14%/wk while MPC/DK/CLMT cleared highs) = crack-spread signal; it was visible in the discovery rows a week before the RSI-78 chase zone.
- Fix (done this session): VLO + PSX added to the energy coverage so the lane exists; follow-up row filed for a proper refining-lane look.
The rest of the board
| Name | 30d | Verdict | Note |
|---|---|---|---|
| PYPL | +33%, RSI 80 | ✅ Not our game (event pop) | The move is the July 15 $53B Stripe/Advent cash takeover bid — a single-day gap no written trigger catches, and the desk captured the bid same-day (it is in the feed). The honest residue is positive: PYPL sat on bargain-bin + lazarus lists cheap enough to draw an acquirer — the value read was validated by a bidder before the tape. Nothing to fix; front-running M&A is not the style. |
| S | +30% | ⚠️ Conviction miss (sector-level) | Cybersec flagged as a rotation-IN sector two weeks running in the rotation map; no single-name trigger existed. Asymmetry worth fixing: the playbook carries cybersec breakdown (exit) triggers but no entry conditions. |
| SOXS, DUST | +55%, +30% | ✅ Not our game | 3x inverse ETFs. Style excludes leverage — but note both are the inverse expressions of calls the desk DID make (the chip derate; the gold breakdown, whose GLD < $410 confirmation fired this week). The views were right; the only instruments that "ran" are ones we don't touch. Clean pass. |
| LCID | +37% | ✅ Not our game | Pre-profit EV turnaround squeeze. Outside trend-hold style. |
| APGE, DAVE, ABVX, ERAS, ACHC, DFTX | +38–81% | ✅ Clean pass | Discovery-net names; the overextension screen classes the cohort as exhaustion (RSI > 70 on sub-1× volume: DAVE 0.79×, ACHC 0.73×). Parabolas on fumes, again. |
| SSPC, SPCQ | +89%, +84% | ✅ Clean pass | Thematic ETF beta, 0.58× volume at RSI 75+. |
Carried from last week — still open
The 2026-07-13 autopsy prescribed an expression ladder for biotech-capital-cycle (a written trigger for the mid-cap leg so the next TXG resolves to a documented yes/no). The ratification gate exists (TMO/DHR/MEDP positive 3m + RSI > 60 together) but the mid-cap entry trigger was never written. This coming week is the checkpoint: DHR (Jul 21), MEDP (Jul 22), TMO + ICLR (Jul 23) all print.
The mirror screen (what got cheap while we watched winners)
The quality-on-sale screen's three survivors: PL (−57% from high, RSI 30, +26% revenue YoY, FCF-positive per EDGAR), WDAY (−42%, RSI 60, +13% YoY, $2.78B FCF), ARM (−41%, RSI 38 — already carries a desk call from the K3 report). PL is the one with no desk look at all — filed as a candidate question, not a view.
Verdict + reasoning
Verdict: 1 coverage miss (refining lane — fixed at the watchlist level this session), 1 conviction miss (S — the recurring "no written trigger" failure), 13 clean passes (PYPL reclassified during the pass: its move is the July 15 $53B takeover bid, captured same-day — an event pop, not a missed entry). The screen keeps validating the style filter: most of what runs +30%/30d is exhaustion-classed leverage and micro-cap beta the desk is right to skip. The failure mode that repeats is never seeing — it's seeing, writing it up, and leaving no entry condition. Two autopsies in a row now name the same fix: when a thesis or watch flags a name, the artifact is not done until it carries either a written trigger or a documented "not our game."
Sources
deno task market -- miss-screen/stretch-scan/outlier-scan/trigger-check, all data-as-of 2026-07-17research/investigations/2026-07-13-miss-autopsy-week-ending-2026-07-10.mdresearch/issues/2026-07-18-market-brief.md(the reader-facing week story)
Related
9 eventsNo direct external sources are attached to this read.