Miss Autopsy — week ending 2026-09-25

Investigation

Miss Autopsy — week ending 2026-09-25

Internal accountability note, prepared September 29 on the September 28 settled screen. This is the required Loop 4 companion to the Monday flagship. It reads the desk's own coverage and decisions; its judgments do not enter the public brief. No new price pull or backtest was made for this note.

What the screen actually says

The already run deno task market -- miss-screen receipt (/private/tmp/tickertape-eod-2026-09-28-miss-screen.log) screened 1,126 priced symbols at the September 28 close. Twenty of 23 movers cleared +30% over 30 days and $10 million of daily dollar volume but had neither a held classification nor an ACTIVE.json active/monitoring key_tickers claim. Three were covered. “MISS” therefore means unclaimed by that join; it does not by itself establish that the desk failed to notice a move, passed an entry, or had an executable signal at the prior close. The September 25 week-close brief and September 21–25 dated scans are the earlier-view controls below. The prior September 21 autopsy already recorded the crypto and BWET ownership gaps; those are continuing, not fresh discoveries.

Screen MISS, 30-day move In our data before Sept 28? Checked source and candidate owner T−1 trigger / judgment
LINK-USD +30% or more Yes; Sept 23 crypto scan and Sept 25 monster scan. Crypto is not claimed by an active perspective. No prior entry rule established in checked files; crypto ownership remains a deliberate open decision, not a proven pass.
AVAX-USD +30% or more Yes; Sept 23 crypto scan and Sept 25 brief. No active perspective claim. Sept 23 brief already showed the move; no prior executable entry rule established.
DOT-USD +30% or more Yes; Sept 23/25 crypto scans; prior autopsy named DOT in the crypto cluster. Same unresolved crypto ownership; no T−1 trigger verified.
ZCSH +30% or more Yes; Sept 21–25 Hot Signals / market-conditions outputs. Candidate crypto/theme coverage. Visibility is proven; entry condition and prior pass are not.
GLNK +30% or more Yes; Sept 22–25 market-conditions and monster outputs. Candidate crypto/theme coverage. Visibility is proven; entry condition and prior pass are not.
AMD +30% or more Yes; Sept 25 week-close brief explicitly described the Monday CPU/semiconductor surge and AMD's rise. Candidate semiconductor/agent-compute owner. The week-close brief recognized the move after it began; no T−1 entry trigger verified here. This is an ownership gap, not a silent coverage miss.
MXL +30% or more Yes; Sept 21/22 brief had a numeric watch; Sept 25 brief reports its +10% day and a 3/5 watch. Candidate semiconductor/optical owner. The written entry_trigger_above changed from $91 on Sept 21/22 to a 20-day-based $71.89 on Sept 25; the Sept 25 price $93.84 crossed the latter, but the advisory still showed no suggestion row. This is an observed watch-to-action gap; its reason needs the decision dossier before calling it a bad pass.
SWKS +30% or more Yes; Sept 21–25 monster scans. Candidate semiconductor owner. A scan presence is not an entry rule; T−1 action level not verified.
TEM +30% or more Yes; Sept 23 AI scan and Sept 21–25 monster scans. Candidate biotech-capital-cycle/AI-health owner. No T−1 action level verified.
TXG +30% or more Yes; Sept 21 brief already listed TXG in stretch exhaustion and Sept 25 Hot Signals showed +33.4% over 30 days. Candidate biotech-capital-cycle owner. Stretch was a risk warning, not a buy trigger; no missed long-entry rule verified.
SDGR +30% or more Yes; Sept 23 entry-zone row and Sept 25 brief's biotech candidate. Candidate biotech-capital-cycle owner. Sept 25 entry zone was $23–$26 while the price was $28.85, 11% above its top; the desk was waiting for a pullback. This is a visible conditional watch, not an unobserved move.
KOD +138% Yes; Sept 21–25 monster scans. Candidate biotech owner requires company evidence. Sept 28 RSI 90 makes an after-the-fact chase especially unsafe; no earlier actionable catalyst/trigger established in checked corpus.
P +39% Yes; Sept 25 week-close brief called out Everpure's +20.99% week and S&P 500 inclusion, and put P atop AI Infrastructure. Candidate AI infrastructure owner. Cause and tape were visible; no prior entry condition verified. This is research ownership rather than discovery failure.
SMTC +33% Yes; Sept 25 optical-supply-chain and monster scans. Candidate optical owner. No T−1 action level verified.
ASST +33% Yes; Sept 21–25 Hot Signals / monster outputs. No active perspective claim. Visibility is proven; business/catalyst basis and T−1 trigger unverified.
SEI +39% Yes; Sept 21–25 monster scans; Sept 28 screen places it on AI-power/SpaceX-supply-chain lists. Candidate power/buildout owner. No T−1 action level verified.
GRAL +66% Yes; Sept 21–25 Hot Signals, but Sept 23 digest explicitly held its stale Sept 22 bar. Candidate biotech owner. That stale-bar hold prevents treating the Sept 23 output as an actionable clean price signal; later T−1 trigger unverified.
IOVA +37% Yes; Sept 21 brief listed stretch exhaustion and Sept 25 Hot Signals showed +36.1% over 30 days. Candidate biotech owner. Stretch was a warning; no missed long-entry rule verified.
BWET +71% Yes; Sept 21 prior autopsy already examined freight and identified missing trigger/ownership routing; Sept 25 Hot Signals carried BWET. Candidate energy/freight owner. Continuing conviction/coverage routing debt, not a new discovery. Prior autopsy's proposed reopen condition was not shown as a fired machine trigger in this bounded check.
COHU +38% Yes; Sept 21–23 monster-discovery and Sept 25 monster scan. Candidate semiconductor-equipment owner. No T−1 action level verified.

Disposition

The immediate failure is ownership of known observations. The screen's 20 names should not be counted as 20 bad trading decisions: only MXL and SDGR had specific dated watch/entry evidence in this bounded check, and SDGR's condition was still unmet on September 25. Crypto and BWET were already called out in the previous autopsy; repeating them without an owner or explicit exclusion keeps the same alert recurring. A single grouped follow-up is warranted for the next research/money-loop review: decide crypto ownership or decline it explicitly; assign semiconductor, biotech and freight perspective membership where a thesis exists; compare the dated MXL/SDGR conditions and the remaining names with actual T−1 decisions before minting any trigger. This note does not silently add 20 entry rules or claim to have settled the trade counterfactuals.

Sources and limits

  • September 28 miss-screen saved terminal receipt: /private/tmp/tickertape-eod-2026-09-28-miss-screen.log; the screen reads priced summaries, paper classifications and active/monitoring perspective key_tickers.
  • Price truth for the screen's September 28 mover rows: research/market-engine/data/summaries/*.json, read by miss-screen through the shared summary store; the saved receipt above provides the printed threshold, counts and named rows.
  • Week-close brief, September 21–25 dated Hot Signals, crypto, monster, AI and optical scans, and the prior Miss Autopsy.
  • No historical execution replay, raw OHLC inspection or individual company deep dives were performed. The +30% entries for clustered symbols reflect the screen threshold; exact per-symbol returns were not printed in the saved text receipt.
10 events
DateTypeEvent
Sep 30 ScanMonster Scan Sep 28 ScanMonster Scan Sep 29 ScanMonster Scan Sep 25 ScanMonster Scan Oct 1 ScanMonster Scan Oct 2 ScanMonster Scan Sep 24 ScanMonster Scan Sep 22 ScanMonster Scan Oct 5 ScanMonster Scan Sep 23 ScanMonster Scan

No direct external sources are attached to this read.