Miss Autopsy — week ending 2026-10-16: three repair names passed at their flags, three found late, five declines still unwritten

Investigation

Miss Autopsy — week ending 2026-10-16: three repair names passed at their flags, three found late, five declines still unwritten

Internal accountability note, written October 19 on the October 19 settled screen (desk date = SPY's October 19 close; the week under review is October 12–16). The screen's 30-day windows run September 18 to October 19 (21 bars; GRAIL 20), so they include Monday's session. Tape figures are engine-computed fields from the desk's summaries at the October 19 close unless dated otherwise; table cells are live bindings. Paper ledgers and session logs, the decision ledger's agent rows and the desk's call records were read. The real-money accounts were not, and the screen's two covered movers, whose held class draws on them, are not unpacked. Nothing was fetched. Tonight's news is thin (the macro pull failed; per-ticker pulls ran on a fallback provider), so a missing cause below is a gap in the pull, not evidence of none.

The short version

  • The screen: 1,289 symbols; 13 movers at +30% or more over 30 days with $10 million or more of daily dollar volume, none skipped as stale. Two are covered, 11 unclaimed, no own-or-decline cluster. Six of the 11 were on the October 12 screen (SHOP, IOVA, KOD, GRAL, NAUT, PACB). Ten tracked blindside receipts.
  • Three conviction misses, one cause. Accenture, Shopify and GRAIL carry written levels. Accenture met its September 25 call on October 1; Shopify closed inside its August 27 deep dive's flag zone in late September. Both times the read applied the one-year rule (flat or down on the year) and called the move a pop or a range: last week's Zscaler conflict, the rule against the user's bottom → pop → flag → breakout, on two more names. It needs his ruling.
  • Three coverage misses. PTC, RXO and XP entered our data only when the market-wide screen added them mid-run (October 14, 14 and 16). None has a news file.
  • Five declines still unwritten. IOVA, KOD, NAUT and PACB are back for a third or fourth autopsy; MAZE is new. Each fails on business quality, a broken long trend, or both.
  • Two corrections to the October 12 note. Shopify has a written entry (the August 27 deep dive: a $138–150 flag, a close above $182.19 to confirm). The energy-to-humanitarian campaign's missing watch-for field is not a gap; the engine ledger withdrew that diagnosis on September 22.
  • Trigger check: one newly crossed row, not zero. The biotech campaign's TXG arm (RSI under 45) crossed with its gate unmet: a documented no under the July rule.
  • The books touched none of the 21 names. Paper fills were stops and trend exits into the S&P 500's worst week since June, three +20% profit targets, and degen's index pair, bought October 13 and sold October 14.

Trigger check (October 19 close)

Campaign (status) Clause State as printed Now Reading
biotech-capital-cycle (monitoring) TXG RSI < 45 FIRED, newly crossed ("3d ago" as printed) 42.5 The July rule's arm. It converts only on TXG RSI back over 55 with price over its 20-day, and only with the gate confirming
biotech-capital-cycle (monitoring) TMO, DHR, MEDP all RSI > 60 ARMED 58.6 / 56.2 / 54 Gate unmet. Danaher, Thermo Fisher and Medpace are scheduled to report October 21 (October 16 brief, provider dates)
biotech-capital-cycle (monitoring) CERT RSI > 55 DORMANT, converted September 15 56.5 Last week's FIRED line now reads correctly after the sequential-clause fix (candidate-fixed October 13, awaiting cold review)
us-energy-dominance (monitoring) USO > $110 FIRED, standing 21+ sessions 160.81 A level, not news

The standing FIRED rows on closed and reopen-trigger campaigns (EWT, GLD, BTC, DBA, USO over $130) are history under the date rule; none reopens. The five legacy-format active campaigns print only unparsed clauses. Energy-to-humanitarian keeps its triggers in its next-review and fails-if fields, read at the October 14 checkpoint (the September CPI test); five other campaigns were reviewed October 16.

The eleven misses

MISS means no book holds the symbol and no active or monitoring campaign claims it. The repeat count includes this autopsy.

Conviction misses: seen, written down, passed

MISS As of 30D (%) 1Y (%) vs 200-day (%) In our data / what the desk wrote T−1 trigger and resolution
ACN 2026-10-19 30.69 -2.17 23 Government-compute and NVIDIA open-agent lists. September 25 watch call: after the October 1 report, a close above $197.63, the top of a seven-week base on the 200-day; invalid below $172.19. Met October 1 at $212.30 (code and agent flags). That night's brief: "this is the pop; the entry is a flag that holds the 200-day (about $196.29)." The retest came October 5–6 at $195.05 and $193.40, 0.2% and 0.9% under the 200-day (washout scans). No desk prose read it; after October 2 Accenture appears only in mechanical rows Its own trigger fired on time; the miss is conversion. A watch call is logged, not acted on, and the one-year rule (down on the year, alpha -16.35) moved the entry in prose. Call ledger since September 25: +34.54% (raw). Resolution: grade at the October 23 review; the rule question goes to the user
SHOP 2026-10-19 33.12 3.86 31.8 Focus, retail and three other lists. August 27 deep dive: "post-print flag $138–150; a breakout above $182 confirms", stop a close under the 200-day. Closes at and inside the zone: $137.92 (September 21, eight cents under its floor), $148.26 (September 29), $149.09 (October 1), read as "A range, not a trend" and "recovering, not yet trending". Read as extended or stretched in four of the five briefs October 12–16 The zone was the trigger and nothing watched it: the deep dive's levels never became a target; the only Shopify target is a July 31 watch-the-floor call, still open. The breakout line has not printed (October 15 close $178.14, 2.22% under its high, per that day's brief). Resolution: register the zone (follow-up 2); a close above $182.19 stays the live trigger; report scheduled November 3
GRAL 2026-10-19 51.57 33.28 60.8 Biotech-scan and discovery lists. October 16 biotech scan, "WATCH: deep week": wait for a flag that holds the 200-day, then a close back over the 20-day; a close under the 200-day voids it. Backdrop: the September 23 FDA advisory vote (GRAIL's investor page, October 19 pull) A written trigger exists and is unmet: 7.4% under its 20-day. Diagnostics sit outside the tools-and-CRO campaign. Resolution: the biotech-scan condition is the row; fourth autopsy

The pattern. All three sit above their 200-day averages. Accenture (down on the year) and Shopify (flat), both with negative alpha, read as unrepaired under the long-trend rule and as flag-then-breakout under the user's pattern. The week cut both ways: Zscaler's return since its September 25 call fell from +16.95% (October 12) to +10.09% (call ledger), Shopify fell 4.55% over the seven days to October 19, Accenture rose 10.27%. Three cases make it a rule question.

Causes. Accenture's run starts with its October 1 fiscal fourth-quarter report: revenue above its guided range (company release, news pulled October 6) and reported EPS of $3.29 against a $3.18 consensus (provider record, basis not stated). This week's leg has no cause on file: news not pulled since October 6, an 8-K (October 14) and the 10-K (October 16) captured but unread, no stock event, one opinion post on X (October 19). Shopify: the October 19 pull holds ETF-flow and quote pages, and the October 12 brief carried a single-source Cantor target raise; cause not established.

Coverage misses: not in our data until the run was mostly done

MISS As of 30D (%) 1Y (%) vs 200-day (%) When it entered our data T−1 trigger and resolution
PTC 2026-10-19 42.07 -6.22 32.2 PTC Inc., application software. Added to the discovery list October 14 by the market-wide screen, already +41.7% over 30 days (October 14 monster scan). The decision ledger's October 14 advisory row: "overbought-stretch (wait-for-the-base)" Nothing on disk could have fired at T−1. Down on the year with negative alpha (-20.4): a repair shape like Accenture's. It held flat through Friday's software rout. Resolution: not our game until a software lane exists; the October 12 commission-or-decline on security software is still open
RXO 2026-10-19 40.67 57.02 39.6 RXO, Inc., US truck brokerage. Added October 14 at +47.4% over 30 days (October 14 monster scan); same advisory row Intact one-year trend. Freight has had no owner since the September 28 autopsy asked for a grouped decision. Resolution: considered, dropped (follow-up 8)
XP 2026-10-19 51.22 73.31 59.4 XP Inc., Brazilian financial services. Added October 16 at +47.9% over 30 days (October 16 monster scan) Intact one-year trend. No thesis covers Brazilian brokerage. Resolution: not our game

All three print the stretch screen's exhaustion class at the October 19 close (rising on 0.65 to 0.79 times average volume), as does Accenture (0.88 times). Cause unknown for all three: no news file, nothing in stock events, and the week's captures hold only a PTC director's Form 4 (October 16, an option exercise). The discovery screen found them late, as built. A coverage add would change nothing: the paper desk refuses discovery-only names (mention-scalper and momentum-trader, October 16), and the missing piece is a thesis.

Seen, declined in practice, never recorded

MISS As of 30D (%) 1Y (%) vs 200-day (%) What is on file Resolution
IOVA 2026-10-19 38.63 512.5 165.4 Discovery list since July 30. 2026 revenue guidance raised to $410–420 million from $350–370 million on September 29 (BioSpace release; news pulled September 30). An October 5 8-K captured, unread Commercial-stage biotech with no thesis. Not our game; fourth autopsy
KOD 2026-10-19 179.27 416.1 133.4 Discovery list. Phase 3 wet-AMD data (Barchart, October 2; news pulled October 14). June-quarter cash $125.9 million with runway into 2027 (company release, August 13) Clinical-stage binary on a short runway. Not our game; fourth autopsy
NAUT 2026-10-19 172.26 136.94 26.4 Biotech-capital-cycle list, not campaign-claimed. Monday's jump on 1.58 times volume: cause unknown. News pulled after Monday's close holds quote pages and an October 27 earnings date; no other stock event; no capture. FY2025 diluted EPS −$0.47 and operating cash flow −$50.7 million (StockTitan, same pull). A September 18 prospectus supplement (424B5) captured, unread. The October 19 board tagged it "bottom pop" Business-quality veto: a sub-$3 loss-maker. Not our game; third autopsy
PACB 2026-10-19 34.31 -7.07 15.3 Biotech-capital-cycle and IBB lists. The window holds the October 5 pop on two diagnostics partners (StockStory per the October 12 autopsy; repeated in the October 19 pull). June-quarter revenue $39.0 million, slightly below a year earlier (company release, August 5) Down on the year with negative alpha (-21.25) and shrinking revenue. Not our game; third autopsy
MAZE 2026-10-19 31.88 -0.17 -7.8 Discovery list; seen all week only on the washout scan ("death cross intact"). Clinical-stage kidney program (company release, August 11). An October 16 8-K captured, unread. News last pulled September 22: not pulled for this move Under its 200-day on a flat year with negative alpha (-14.35). Not our game; first appearance

Tracked blindside receipts (October 19)

Symbol As of 1D (%) 7D (%) 1Y (%) vs 200-day (%) Owner on file Cause check Disposition
AGIO 2026-10-19 11.43 13.24 -18.17 11.5 August 24 watch call: invalidated October 12 (verified close under the 200-day), "tape broken" on the board October 14, in zone by code October 19, all on one open call News pulled October 19: an H.C. Wainwright target raise (StockAnalysis); report scheduled October 29; no capture. Cause not established Owned; invalidation recorded the evening of October 19 after the bars confirmed closes of $30.82, $30.21 and $30.97 on October 12-14, under the 200-day (about $31.21)
ALMU 2026-10-19 -1.89 -25.14 -52.15 -54.8 July 1 deep dive; profile view "no view"; the commissioning board declined a new dive (one source family) News pulled October 19, newest item September 25; an October 14 Form 4 and an October 15 trade-press capture. Cause unknown Owned; broken; no action
AXTI 2026-10-19 10.34 -8.37 1056.62 -10.2 The July 8 pass, review date October 1, unreviewed October 19 pull holds aggregator pages; report scheduled October 29. Cause unknown Owned by a lapsed pass; long trend under test, not broken
BRBR 2026-10-19 1.7 11.64 -70.35 -31.4 September 28 profile; thesis rated broken News last pulled October 9: not pulled for this move Owned; a bounce in a broken chart
BRUN 2026-10-19 -1.24 -25.53 -29.87 -51.9 None (AI-infrastructure list) October 19 pull: newest item an October 6 contract; captures hold promoter posts. The GPU-host selloff is sourced (CoreWeave's default swaps, October 16 brief); Boost Run's own cause is not Unowned; broken; decline
CIFR 2026-10-19 -0.45 -31.23 -55.1 -50.7 The July 8 pass (reopen needs de-leveraging); the neocloud focus-lane read $850 million zero-coupon convertible and share exchange (8-Ks October 13, 14 and 16 captured; summarized in the October 16 brief and the October 19 pull) Owned; the pass's own condition moved the wrong way
CLF 2026-10-19 16.34 9.39 -12.86 24.4 May 4 deep dive (history: it predates the current frame) The October 19 report: direction sourced, figures third-party and unreconciled (stock event); 8-K captured, unread Reconcile from the 8-K before any read
ESLT 2026-10-19 0.53 16.16 54.79 -2.8 None: on the drone-defense list, which the drone campaign does not claim (it names AVAV, KTOS, RCAT) News pulled the evening of October 15, newest item from August; an October 19 6-K captured, unread. Cause unknown Unowned; long trend under test; read the 6-K
NAUT 2026-10-19 27.67 21.2 136.94 26.4 None Above Not our game
VICR 2026-10-19 -0.43 -26.49 228.53 -15.1 ai-power-delivery key name; the July 31 stop at $178 is open Cause of the October 16 drop unknown, per the October 16 campaign review; report scheduled October 20 before the open Owned; the report is the campaign's first reported test

Seven of ten have something on file; only Vicor has a live thesis owner. Boost Run, Elbit and Nautilus have none.

Not misses

  • ATKR: acquired by Prysmian, closed October 19 at $95.00 cash (SEC 8-K, accession 0001666138-26-000041), and retired from coverage that night. It has no summary row and is not on the screen.
  • CLF: the October 19 print figures on disk are third-party and unreconciled; the stock event says so and nothing here uses them.

What the desk got right this week

Dated flags that fired and were borne out:

  • The stretch screen's October 12 exhaustion class. All six names fell over the seven days to October 19: P −18.17%, BWET −13.73%, KOD −7.97%, NTSK −5.91%, LIND −5.27%, IOVA −2.40%, against SPY −1.49% and QQQ −3.71% (desk summaries).
  • CoreWeave's September 25 watch call. Its invalidation (a close under $85.64) fired September 29 and the desk stayed out; CoreWeave fell 25.21% in the week to October 16 (October 16 brief), −30.22% since the call (call ledger).
  • Cipher's July 8 pass. It required de-leveraging; Cipher priced an $850 million convertible and fell 33.58% in the week to October 16 (October 16 brief).
  • Lumentum (October 2) and Constellation (October 9) watch calls. Both invalidations fired October 16, before any entry flag. Since their calls: −23.12% and −11.70% (call ledger, October 19).
  • Shopify's "don't chase" reads (October 13 and 15 desk reads, October 15 brief). It fell 4.55% over the seven days to October 19. Right on the week, wrong on the month.
  • The October 12 note's SKYY warning. It said the parked proposal contradicted its own break line. The book filled it under that line and the stop hit four sessions later.

Wrong, lapsed, or still open from October 12

  • The refiner passes. Calumet and Delek, passed July 18 as extended, closed their 90-day windows on October 16 at +30.09 and +15.64 points over SPY (call ledger). The same note's Marathon, Phillips 66 and Valero watch calls finished +48.77, +37.56 and +39.09 points (logged, unscored). That is the price of "RSI over 65 without a prior reset is a documented pass".
  • AGIO, 25 days past its September 24 review, printed "watch (in zone)" in the October 19 brief's mechanical desk-call block a week after its own invalidation. The evening session recorded the invalidation (bars: closes under the 200-day on October 12-14) and the block is rebuilt from the flags, so the published brief is checked for it (follow-up 3).
  • SDGR: still no own-or-decline. It fell through its $23–26 zone to $22.60 (October 16 target review) and closed at $23.12 on October 19 (desk summaries), at the zone's floor with the $20.50 stop untouched. The silence cost nothing this week; the decision is still owed.
  • Declines: none recorded; IOVA, KOD, NAUT and PACB returned.
  • Security software: partly answered. The October 16 article wrote a fund-level call on CIBR (a flag that holds its $81.59 200-day, then two closes over the 50-day). No ruling on Zscaler is on file.
  • TXG: the arm fired; the campaign log has no entry since October 12 saying what it means.
  • Lifecycles: the AXT pass, review date October 1, is unreviewed for a third autopsy; Okta's completed pass still reads open.

What the books did

From the paper ledgers and session logs.

  • No paper book traded any of the 11 misses or the 10 blindside names. Six misses sit only on the discovery list, which the paper desk refuses.
  • degen: the Evolved Rule 2 admissions filled October 13 (TQQQ $78.13, SPXL $295.69); the exit check sold both October 14 on "trend-died" (TQQQ $74.12, SPXL $287.63), the day of the hot September CPI. Two of the four fills that trigger the rule's re-grade.
  • swing-kid: SKYY filled October 12 at $172.47, under the book's own $173.15 failed-breakout line, and the 5% stop sold it at $158.51 on October 16 (−8.1% per the exit check). The replay still does not re-check a proposal's break line.
  • Stops and trend exits: claude-trader's Broadcom on its accelerating-loss rule ($328.75, −12.7%, October 16); autotrade's SMH ($545.57, October 15) and SKYY ($160.03, October 19). core70's AMAT stop (−16.9%) and same-day spec rebuy is already registered (D224, default-exit-rules-on-a-spec-arm).
  • Profit targets on the receivers: bench-signals' XLE (+20.4%, October 13), yolo's Gilead (+20.7%, October 15) and Chevron (+20.1%, October 16).
  • Gates held: momentum-trader's market gate stayed shut an eighteenth session; mention-scalper passed all three Friday candidates.

Follow-ups

  1. This note's verdict line (one line, so the screen prints it beside repeats): "open; conviction misses ACN SHOP GRAL; coverage misses PTC RXO XP; not our game IOVA KOD NAUT PACB MAZE".
  2. ops/tasks/TASKS-ENGINE.md: a deep dive's entry zone and stop never became a watched target (Shopify, August 27; closes inside the zone on September 29 and October 1 went unwatched). First check whether any path from deep-dive frontmatter to target review exists; the July 31 Vicor targets came from a company profile's action matrix.
  3. Does a call survive its own verified invalidation? Settled in practice for AGIO, WDAY and TLT on the evening of October 19: the flag receipts record invalidation (AGIO and WDAY checked against the bars). Open question: the code raised each as "in zone" while the invalidation clause is prose that only an agent reads, so a call can print "watch (in zone)" until an agent records the verdict. Check the rebuilt October 19 brief's desk-call block before it publishes; if any of the three still prints in zone, file the DEFECTS.md row.
  4. Biotech campaign log (deno task research:campaign): record that the TXG arm without the gate is a no under the July rule; re-check the gate after the October 21 reports.
  5. Close lifecycles: the AXT pass (review October 1), the Okta pass (September 15), the AGIO call (September 24); grade Accenture at its October 23 review.
  6. The user's call, not filed: on a repair name above its 200-day, does the one-year rule override a met written condition or a deep-dive flag zone? Cases on file: Zscaler, Accenture, Shopify.
  7. SDGR own-or-decline: third ask; the user's call.
  8. Considered, dropped: coverage rows for PTC, RXO and XP (the screen already covers them; the missing piece is a thesis, and the paper desk refuses discovery-only names). A freight own-or-decline for RXO (one name, no cause on file; reopen if a second freight name screens). Reading MAZE's 8-K (clinical-stage, under its 200-day; low value). A counterfactual on Accenture's flag: the retest closed under the 200-day on both days, so whether it "held" is ambiguous and the data cannot settle it.

Sources

  • Screens (October 19 receipts): miss-screen, trigger-check, stretch-scan, outlier-scan, trip-report.
  • Prices: research/market-engine/data/summaries/*.json, October 19 settled close; scan and brief figures as dated in text.
  • Desk reads: research/issues/2026-10-1{2,3,4,5,6}-market-brief.md above the regeneration fence plus their desk-call and lane-tape rows; research/issues/2026-10-19-market-brief.md (mechanical, desk read pending); research/issues/2026-09-21,2026-09-29,2026-10-01,2026-10-02,2026-10-09-market-brief.md for the Shopify and Accenture lines; research/scans/2026-10-0{5,6}-washout-scan.md, research/scans/2026-10-1{2..9}-washout-scan.md, research/scans/2026-10-1{4,6,9}-monster-scan.md, research/scans/2026-10-1{4,6}-monster-discover.md, research/scans/2026-10-16-biotech-scan.md; research/issues/2026-09-25-bottom-feeding-ai-software-split.md; research/issues/2026-10-16-money-has-left-ai-for-oil-and-staples-since-october-8.md.
  • Calls, targets, decisions: research/api-data/desk-calls.json (generated 2026-10-19T22:53Z); research/trading/decisions/CALL-FLAGS.json; research/trading/decisions/LEDGER.json (agent rows since October 12; human-actor rows skipped unread); research/targets/2026-07-31-vicr-target-{entry-zone,stop-loss,profit-target}.md, research/targets/2026-07-31-shop-target-desk-call.md, research/targets/2026-08-24-agio-target-desk-call.md, research/targets/2026-09-23-sdgr-target-entry-zone.md.
  • Research and campaigns: research/deep-dives/2026-08-27-shop-deep-dive.md; research/perspectives/ACTIVE.json (claimed names; campaign reviews October 14–16); research/perspectives/2026-06-04-biotech-capital-cycle/log.md; research/workflow/COMMISSIONING.md (generated 2026-10-20T01:10Z); owner classes from research/mover-attribution.ts.
  • Cause checks: research/market-engine/data/stocks/<SYM>/news/latest.json for all 21 names (PTC, RXO and XP have none); research/stock-events/ (2026-10-01-acn, 2026-10-19-clf, 2026-10-20-vicr, 2026-10-27-naut, 2026-10-29-agio, 2026-10-29-axti earnings); research/capture/{articles,tweets}/ dated October 12–19; research/capture/filings/ since September 18.
  • Coverage and identity: research/market-engine/watchlists/CHANGELOG.jsonl (PTC and RXO added October 14, XP October 16, ATKR retired); research/market-engine/data/stocks/{PTC,RXO,XP}/info.json.
  • Books: research/trading/paper/*/LEDGER.json fills October 12–19; research/trading/paper/{mention-scalper,momentum-trader,swing-kid}/PLAYBOOK.md October 16 sessions; research/trading/TRADING-PATTERNS.md.
  • Engine ledger: ops/tasks/TASKS-ENGINE.md (the withdrawn watch-for row, September 22; the sequential-clause row, candidate-fixed October 13).
  • Primary filings: ATKR 8-K, October 19; CLF 8-K, October 19; ESLT 6-K, October 19; MAZE 8-K, October 16; ACN 8-K, October 14; NAUT 424B5, September 18. Captured as filing stubs; bodies not read.
  • Prior autopsy: week ending October 9.
  • Limits: no fetch, raw OHLC read, execution replay or counterfactual P&L. The two covered movers are not unpacked.
8 events