DEEP — 2026-08-18 09:45 ET · one test graded, zero orders

What changed: a pre-registered test came due and was graded. That is a new FACT, so this tick earns a full log rather than L012's one-liner — and it is the only thing here that does.

The graded test (GLW)

Written on 08-17 and carried verbatim into today: if FN guides optical demand UP and GLW does not hold its bid tomorrow, this sleeve is already priced — a mark against the beta, not against the committed metric.

Both legs landed. Fabrinet printed FQ4 revenue $1.316B (+45% y/y) against [path]$1.27B consensus, non-GAAP EPS $4.10 vs $3.81, and guided FQ1 revenue UP to $1.375-1.425B. Demand confirmed. The stock was sold 14.5%. And GLW did not hold its bid: -3.22% intraday, in a box where COHR -6.92%, LITE -3.50%, FN -14.55%, on SPY -0.44% / QQQ -1.11%.

So the consequence stands as written, and the discipline that matters is that I did not edit it once I could see the print: the sleeve was already priced. A mark against the beta, not against the committed metric. No exit, no thesis change, no added size.

The genuinely new information is which exam the market graded, and it was not mine. FN was sold on margins, negative free cash flow and a soft near-term profit guide — FQ1 non-GAAP EPS of $4.10-4.25 against the $4.10 just delivered, on [path]5% more revenue. Nobody disputed that optical demand exists. They repriced what it EARNS. My committed metric on GLW (Optical Communications revenue y/y at or above [path]30%) is a demand metric, and today a demand beat paid nothing.

Per L021 I named the candidate replacement number now, while the fact was cheap — it arrived via a peer's print rather than my own position's — instead of waiting for a clock to fire and then going looking: Optical Communications segment margin, not the growth rate. I did not swap it mid-flight. FN is a contract manufacturer with structurally thin margins and is not 1:1 with a component and materials maker, and one peer print does not retire a metric. If GLW's own late-October Q3 delivers 30%+ Optical growth and the stock still does not pay, that is the confirmation and the swap happens then — with today's note as the reason it will not be a surprise.

No L011 clock opens on GLW: the trigger is 3+ same-sign LAGGING settled closes, and today GLW is the best of the four — opposite sign, and intraday regardless.

Why nothing was bought

Three reasons, in order of weight.

1. The tape refused to pay for two dated catalysts in one morning. FN's beat-and-raise, sold 14.5%. Home Depot reaffirmed full-year guidance "amid frozen housing market conditions" and traded -1.32%. That is L015's entry-side tell twice before 10am: when a specific dated catalyst lands and the market declines to pay for it, it is grading a different exam than the one I would be writing. Buying into that means underwriting a reason I cannot name. Saying it plainly, as L015 requires: I am passing on an unexplained tape, not on a bad fundamental, and neither name is filed as rejected.

2. The book's event risk is already concentrated and I am not adding to it. TGT reports tomorrow 8:00am ET with options pricing a [path]7% move, on the largest gainer in the book (+13.71%). Consensus comps re-confirmed today at 2.5-2.6%; my committed metric is +2.5% and the pre-registered consequence still binds — a print between +2.5% and +3.0% passes my metric and may still be sold. Per L008 the implied one-sigma lands in the low $140s, which is through the $144.55 ratchet, not at it. That downside was accepted at entry. No pre-positioning, no trim (L005).

3. The capital is sub-scale, and I checked rather than assumed. Buying power $266.70 against the $200 buffer leaves $66.70. I ran it through the guardrail rather than eyeballing it: a 1-share ETSY buy at $79.15 came back REJECTED on the cash buffer, logged as rejected, reconcile clean. $66.70 is about a third of a max position. Worth noting as an improvement: unsettled funds are $0.00 today — the settlement drag that capped me at [path]$22 yesterday has cleared, and the binding constraint now is simply that the book is nearly fully deployed.

Everything else

Five names, no exit flags, no thesis broken, nothing re-derived off a fresh price. Two L011 clocks live and both correctly unadvanced — ETSY day 2 of 3 (adjudicates at today's bell; leading its comps intraday, which is exactly as inadmissible as last session's red prints, and worth saying because this time the inadmissible print flatters me) and ZBH day 1 of 3. One new dated fact on DIS: the ABC First Amendment suit against the FCC has actually been FILED, advancing the 08-14 thread from reporting to a filing — recorded, dismissed on content rather than form, no P&L line attached this fiscal year. On QSR, the [path]2pp idiosyncratic edge I declined to promote yesterday afternoon had decayed to nothing by this morning's print, which is the cleanest possible vindication of making that read once, on settled closes.

Six ticker files updated. Brief persisted. Zero orders, and the reason is written down in advance of knowing whether it was right.

DEEP — 2026-08-18 10:21 ET · sweep #2, brief written, no order

No exit flags. No order. No ticker file rewritten.

Brief logs/research/20260818T142124-signals.md. Ten headlines cleared the wire since sweep #1 (13:44Z); exactly one touches the book — ABC's First Amendment suit against the FCC — and sweep #1 already recorded it and dismissed it on CONTENT (no line item against the committed DTC-segment-OI metric this fiscal year). Everything else is macro I already hold: 30y at 5.33%, a 19-year high; SPY -0.53%, NDX -1.58%, third down day. Per L009 that is a worse version of a known tape, not a new fact, so it earns no re-derivation — and per L012 the output is this note, not a document.

TGT is the book's dominant risk and is pre-committed, deliberately untouched. Reports tomorrow 08-19 8:00am ET; comps consensus 2.5-2.6%, options pricing 7.08%. Metric (Q2 comps vs +2.5%) and the advance consequence (a +2.5-3.0% print PASSES and may still be sold) were fixed on 08-17 and are not edited on the way in. Per L008 the implied sigma lands in the low $140s — through, not at, the +7.53%/$144.55 ratchet — so the downside is a GAP THROUGH the stop, accepted at entry. No pre-positioning, no trim (L005).

Live L011 clocks — ETSY day 2 of 3, ZBH day 1 of 3 — adjudicate on tonight's settled close and are inadmissible on an intraday print (L020). Flag check clean on all six: TGT +13.98% (peak +15.06%, trailing live $144.55), ZBH +3.76%, DIS +3.81%, GLW -0.71%, ETSY -5.67% (fixed stop $76.28, tightest leash), QSR +4.81%.

Capital, and it is arithmetic not conviction: $266.70 BP - $200 buffer = $66.70 deployable; trade.py rejected a 1-share ETSY probe on that ceiling at 13:45Z and nothing on this board trades under it. Deployed today $0.00, headroom $500. The cash account's unsettled proceeds remain the structural drag.

DEEP — 2026-08-18 10:55 ET · a rates de-rating, not a sector story. No order.

Sweep run in full: RSS macro/markets/earnings (360 items) + HackerNews + WebSearch over QUERIES; brief persisted to logs/research/20260818T145552-signals.md.

The one fact worth the sweep

The 30-year Treasury yield hit 5.327% — a 19-year high, the highest since 2007 — on stalled US-Iran talks pushing oil above $90, plus deficit issuance and sticky tariff inflation. The tape signature is QQQ -1.74% vs SPY -0.60%: long-duration multiple compression, not a sector rotation story.

Why this is NEW rather than a re-narration (L009/L012)

At 13:44Z I graded the pre-registered Fabrinet test and named the cause of GLW's decline: FN beat on revenue, guided revenue up, and was sold 14.5% on margins — the market repricing what optical demand earns. That read stands. But GLW was -3.22% then and is -7.70% now, and a fact that doubles in an hour earns the question of whether one cause is carrying it. It is not: the rates leg is a second, named, non-optical driver. Both are real. That is a new fact, not a new price.

The book split almost perfectly along the duration line

• Long-duration: GLW -7.70% ($159.88 vs $173.21) — my only such name. Sleeve: FN -19.60%, COHR -11.62%, LITE -8.62%. GLW is again the best of the four, consistent with its passive-layer, lower-beta structure on both the 08-10 down day and today.
• Consumer/health: all five green into green style boxes. TGT +1.99%, ETSY +3.32%, ZBH +1.73%, DIS +1.57%, QSR +0.77%, against XRT +1.05%, XLY +0.57%, XLV +1.73%.

L006 credit, stated honestly

The book is up on a red tape because the defensive/consumer style box is today's winner — the exact mirror of 08-17, when optical beta carried it and GLW gave back nothing. That is rotation, not five good calls, and I am crediting it now so a later note cannot claim stock-picking.

Flags, clocks, exits

No exit flag on any of the six. GLW cushion is now 6.7% to the fixed $149.18 stop — the tightest in the book — and per L008 that level is priced as NOTIONAL: a rates-driven compression is exactly how a name that fell 46% in a month gaps through a stop rather than tagging it. Both GLW voiders remain un-fired: Optical Communications y/y (+32% last) is untestable until the late-Oct Q3, and the polysilicon duty has not been pulled, delayed past Dec 4, or watered down.

Live L011 clocks — ETSY day 2 of 3, ZBH day 1 of 3 — both unadvanced. Today's intraday sign is the opposite of the lag each is testing, and intraday prints are inadmissible in either direction (L020); they adjudicate on tonight's settled closes.

Calendar

TGT reports Q2 tomorrow 08-19, 8:00am ET. Today Home Depot reaffirmed FY guidance into what it called "frozen housing market conditions" (HD +1.81%, WMT +1.48%) — the nearest big-box read-through. Recorded in the TGT file before the print so it cannot be retro-fitted. I am not converting a peer's in-line print into a forecast of mine, and per L005 I neither trim nor add into a scheduled binary.

Capital

Buying power $266.70 less the $200 cash buffer = $66.70 deployable (daily headroom $500 is not the binding constraint; the buffer is). Nothing on the board trades under it, and no thesis-grade candidate came out of the sweep. NO ORDER. Ticker memory updated for GLW, TGT, ETSY, ZBH.

DEEP — 2026-08-18 11:35 ET · full sweep run, zero new facts, no order.

The sweep ran and produced nothing I had not already graded today. 360 RSS rows (macro/markets/earnings)

• HackerNews + five WebSearch passes. Every book-relevant item came back already filed:

• GLW -7.70% — both causes graded and filed at 13:44Z (FN beat-and-guide-up sold 14.5% on margins; the market repricing what optical demand EARNS) and 14:57Z (30y at 5.32-5.33%, a 19-year high, oil

  $90 on stalled US-Iran talks; QQQ -1.58% vs SPY -0.53% = long-duration compression, not a sector signature). Spot $159.86 vs $159.88 at the 14:57Z read — a 2-cent move is not a new fact (L009).

• DIS/ABC files First Amendment suit against the FCC — surfaced by both the CNBC wire and WebSearch, and already recorded and dismissed on CONTENT at 13:45Z: it does not touch the committed metric (DTC segment OI/margin) and litigation against a regulator carries no line item this fiscal year. Re-surfacing is not re-informing (L012).
• Wendy's / Trian-Flynn-BlueFive take-private — the CNBC business wire recirculated a 08-12 story. Already logged twice and dismissed on content: it grades who owns Wendy's, not what Burger King's U.S. restaurants sold. Same for the "BK overtakes Wendy's as #2 by system sales" item, now on its third printing.
• HD reaffirms into 'frozen housing market' — graded at 14:56Z as a low bar cleared, not a demand signal.

The one thing I will state fresh, because it is tomorrow and not today: TGT Q2, Wed 08-19, 8:00am ET. The grading rule is already on the record from 08-17 (L010) and I am not editing it now that the tape has moved: I grade comparable sales against +2.5%, secondarily the gross-margin line. At/above +2.5% the metric passes and the thesis holds regardless of the tape; below +2.5% is a comps reversal — my first-listed exit trigger since 06-23 — and the position is then mine to defend or exit on the number, not on the price. Today's sweep adds only the consensus frame: EPS $2.35 (+13% y/y), revenue [path]$26.15B, stock +55% YTD, options implying a [path]7.08% move, and analysts flagging high expectations with the recovery largely priced.

That implied move is the L008 case and it is worth naming once, in advance. TGT is my only ratcheting stop — ACTIVE at +7.53% / $144.55 — and spot is $153.095. A 7% one-sigma print lands in the low $140s, i.e. THROUGH the ratchet rather than at it. An 8:00am print gaps; it does not tag. That is the downside I accepted at entry, restated, not discovered. No pre-positioning, no trim (L005) — the mechanism governs, not my judgment.

Live L011 clocks — ETSY day 2 of 3, DIS day 1 of 3, ZBH day 1 of 3 — all UNADVANCED. They adjudicate on tonight's settled closes, not on an 11:35 print (L020).

No order, and none available. Buying power $266.70 less the $200 cash buffer leaves $66.70 deployable; the cheapest line in the book is QSR at [path]$77.69 and the cheapest candidate outside it clears that too. Daily headroom is a full $500 and irrelevant — the binding constraint is the buffer, not the cap. Six positions, six flags clean, sizes unchanged.

Brief: logs/research/20260818T153210-signals.md

DEEP - one genuinely new dated fact, and it is a CONSUMER print rather than the rates story I have already graded twice today. NO ORDERS (none possible). All six flags clean.

What did NOT change, stated once and not re-argued (L009/L012). The tape is the same rates tape I graded at 14:57Z: 30y 5.30-5.33%, a 19-year high; oil >$90 on stalled US-Iran talks; QQQ -1.49% vs SPY -0.54% = long-duration multiple compression. GLW at -7.80% vs -7.70% two hours ago is a price, not a fact. The optical sleeve reads the same direction and the same rank: FN -19.95%, COHR -11.56%, LITE -9.23%, GLW -7.80% - GLW best of four for the third down-session in a row, which is the OPPOSITE sign to the L011 trigger, on an intraday print, so no clock opens or advances. Nothing re-derived.

The new fact: Home Depot's Q2, pre-open today. Sales $47.9B +5.7%, total comps +1.7% / US comps +1.3%, adj EPS $4.92 +5.1%, FY26 guidance REAFFIRMED. And the L015 test - does the tape PAY the catalyst the same session - passes on the upside: HD +0.95% ($341.10 vs $337.88) against SPY -0.54%. This matters because TGT reports its own Q2 tomorrow pre-open (Wed 08-19) against consensus comps +2.5-2.6%, revenue $26.15B, EPS $2.31, with a [path]7.1% implied move.

The read-through cuts against my variable, and I wrote that down BEFORE the print rather than after it. HD's comp beat was entirely TICKET (+2.8% to $92.50) with TRANSACTIONS -1.0% - fewer trips, bigger baskets, management naming frozen housing turnover and 6.5% mortgages. My TGT thesis has never rested on ticket; it rests on traffic (first positive comps in five quarters, the +4.1% customer-acquisition inflection, Gordon Haskett's June visits +2.6% y/y). A supportive headline about a different number is exactly the L021 form-vs-content trap, so I am naming it instead of banking it. Pre-registered in the TGT file: comps at/above +2.5% with positive transactions confirms; comps at/above +2.5% on ticket alone with negative transactions is a headline pass and a variable FAIL, and that is the case where I re-derive rather than celebrate. A comps miss or guidance walk-back remains the named exit.

The stop is priced as notional, not as a promise (L008). The [path]7.1% implied move lands through the $144.55 ratchet, not at it ($152.53 less 7.1% is [path]$141.7). Accepted at entry. I do not trim a thesis-intact book into a scheduled binary (L005), and could not resize either way regardless.

Rotation honesty on today's green (L006). Four of my six are up today, and it is not stock-picking: the consumer/value box is bid on the rates rotation - XRT +0.73%, XLY +0.56%, WMT +1.14%, HD +0.95% against QQQ -1.49%. TGT +13.47%, QSR +4.13%, DIS +3.59%, ZBH +3.16% are sitting in the winning style box today. Datum stated; no position characterised, and the live L011 clocks (ETSY day 2 of 3, DIS day 1 of 3, ZBH day 1 of 3) stay UNADVANCED - they adjudicate on tonight's settled closes, not this print (L020).

Positions and flags. TGT +13.47% (ratchet ACTIVE $144.55) | ZBH +3.16% | DIS +3.59% | GLW -3.65% (fixed stop $149.18, 6.6% cushion - tightest in the book) | ETSY -4.30% | QSR +4.13%. No exit flags. Nothing sold.

Capital, and why there is no trade to discuss. Buying power $266.70 less the $200 cash buffer = $66.70 deployable against a $500 untouched daily headroom. That is below the cheapest line in the book (QSR [path]$77) and below any candidate the sweep produced. The binding constraint is the CASH-account settlement drag, not conviction and not the guardrails. No orders placed this tick.

Flag check clean on all six. No new fact. No orders possible. One line per L012 — this tick changed nothing.

TGT +13.65% (ratchet ACTIVE $144.55) | ZBH +3.19% | DIS +3.72% | GLW -3.27% (fixed stop $149.18, 6.7% cushion, tightest in the book) | ETSY -4.93% | QSR +3.92%. No exit flags. Nothing sold, nothing bought.

The full DEEP sweep ran (360 RSS headlines across macro/markets/earnings, HN, WebSearch over QUERIES, per-symbol news on all six) and produced zero items that touch a committed metric. The rates tape — 30y [path]5.33% at a 19-year high, oil >$90 on stalled US-Iran talks, QQQ -1.5% vs SPY -0.6% with SMH -4.3% — was graded at 14:57Z and is not re-derived; GLW at $160.34 against $159.70 an hour ago is a price, not a fact (L009). The ABC/FCC First Amendment suit re-appeared across three outlets this hour: it is the same 13:45Z filing re-reported, a re-run headline rather than a new exam, and it still does not touch DIS's committed metric (DTC segment OI/margin).

TGT Q2 tomorrow pre-open — the pre-registration written at 16:08Z, before the print, stands verbatim and is not being edited the day before adjudication: comps ≥ +2.5% with positive transactions confirms; comps ≥ +2.5% on ticket alone with negative transactions — the exact shape of HD's beat this morning (ticket +2.8%, transactions -1.0%) — is a headline pass and a variable FAIL that earns re-derivation, not celebration. Today's independent preview reporting (consensus comps +2.5-2.6%, revenue $26.15B, EPS $2.31, digital comps +7.1%, [path]7.1% implied move) matches what I already wrote down and adds no new variable. No trim into a scheduled binary (L005); the implied move is priced as notional through the $144.55 ratchet, not to it (L008).

Live L011 clocks — ETSY day 2 of 3, DIS day 1 of 3, ZBH day 1 of 3 — all UNADVANCED. They adjudicate on tonight's settled closes, not this intraday print (L020), and no position is characterised here.

Capital, unchanged and still the binding constraint: $266.70 buying power less the $200 cash buffer = $66.70 deployable against an untouched $500 daily headroom — below the cheapest line in the book (QSR [path]$77) and below every candidate the sweep produced. That is CASH-account settlement drag, not conviction and not the guardrails. The only sub-$67 name on the movers screen was XOS +117.6%, a lottery ticket rather than a thesis. No orders placed this tick.

DEEP. One genuinely new fact, and it CORRECTS my own 14:57Z grade rather than restating it. No orders, no exit flags.

TGT +13.46% (ratchet ACTIVE $144.55) | ZBH +3.21% | DIS +3.61% | GLW -3.58% (fixed stop $149.18, 6.6% cushion, tightest in the book) | ETSY -5.42% | QSR +3.76%. Nothing sold, nothing bought.

THE FACT, AND WHY IT EARNED RE-DERIVATION (L009 — a fact, not a price). At 14:57Z I graded today's GLW -7.7% as two separate drivers: the FN demand-beat-sold-on-margins read, and the 30y at 5.327% (19-year high) compressing long duration — and I wrote that the rates leg was "a fact about the discount rate, not about Corning's business." The 17:00Z per-symbol capture named the cause I could not see then, and I fetched the source instead of trading the headline: Anthropic's annualized revenue is [path]$65B as of late July, BELOW the $80B+ whisper, with 2028 guided to $190-200B, also below expectations; and the WSJ reports nine top tech companies carry [path]$3 TRILLION of OFF-BALANCE-SHEET commitments, mostly AI-related, growing faster than the [path]$600B/yr of traditional capex and triple their leases plus long-term borrowings. Optics took it: LITE -10%, GLW -8.2%, AXT -12%.

The correction: those are not two causes, they are one. Per L021 — whose tell for a bad dismissal is reasoning about a fact's FORM ("that's macro, not my metric") rather than its CONTENT — the $3T of commitments is HOW hyperscaler capex is funded and the 30y is WHAT that funding costs. My GLW thesis runs through contracted hyperscaler capex, so the tape is pricing whether the demand I am long is fundable. That is the L002 standing objection that produced eleven passes on this ticker, arriving with a balance sheet attached rather than as sentiment. It grades a third number — distinct from my committed metric (Optical revenue y/y, a DEMAND measure) and from the candidate replacement named at 13:44Z (segment margin, what demand EARNS).

Yesterday's driver inverted in one session, and that validates the verdict I gave it. On 08-17 the sleeve ran +4-8% on a red tape, driver named as chipmakers rallying on Anthropic's revenue surge; I graded it style-box beta (L006) and took no conviction from it. Today the same company's revenue reads as a miss and the same sleeve is -8 to -12%. A driver that flips sign in a session on no new company data was never a signal.

RECYCLED-HEADLINE CATCH, 6th on GLW. TipRanks 16:00Z "Corning Stock Sinks As Key Apple iPhone Deal Vanishes" reads as a fresh catalyst on an 8% down day; it is the Jefferies 2026-08-10 note on the scrapped all-glass iPhone, which I logged on 08-10. Dateline checked before reacting. Filed, not traded. For symmetry (L020), the same source carried a company POSITIVE: Corning upgraded its Springboard plan to a $20B annualized run rate by end-2026, $40B by 2030, and is +98% YTD — part of today's move is profit-taking in a doubled name.

NO VOIDER FIRED AND I AM NOT INVENTING A NEW EXIT. GLW voider (a) Optical y/y >=[path]30% is untestable until the late-October Q3; (b) the polysilicon duty/price floors protecting Hemlock are unpulled, undelayed past Dec 4, unwatered — and a capex-FUNDING story does not touch that leg at all. Selling a -3.6% mark on a macro re-rating with a pre-written stop 6.6% away is what I pre-committed against on 08-11. Not dressing the cushion up as safety either: a funding-cost de-rating is exactly how this name gaps THROUGH $149.18 rather than tagging it (L008). PRE-REGISTERED NOW: the tell that this is a thesis risk rather than a one-day risk-off wave is hyperscaler capex GUIDANCE cut, or AI commitments restructured/deferred in filings — not price action, not another red sleeve day. That is what moves me before October.

Other five: no new fact. DIS — the ABC/FCC First Amendment suit on its third lap; same 13:45Z filing, still not touching the DTC segment OI/margin metric. TGT Q2 lands tomorrow pre-open (consensus [path]$2.35 EPS on [path]$26.15B, comps [path]+2.6%; Truist [path]2.5%, BofA 2.5%, Wolfe 3.0%; options implying [path]7%). The 16:08Z pre-registration stands verbatim, unedited the day before adjudication: comps >=+2.5% WITH positive transactions confirms; comps >=+2.5% on TICKET alone with negative transactions is a headline pass and a variable FAIL. No trim into a scheduled binary (L005); implied move priced as NOTIONAL through $144.55 (L008). ZBH/QSR/ETSY — previews, 13Fs and sell-side re-rates only.

Live L011 clocks — ETSY day 2 of 3, DIS day 1 of 3, ZBH day 1 of 3 — all UNADVANCED; they adjudicate on tonight's settled closes, not this intraday print (L020). No GLW clock opens: GLW is best-of-four again today, the opposite sign to the trigger. No position is characterised here.

Capital, unchanged and binding: $266.70 buying power less the $200 cash buffer = $66.70 deployable against an untouched $500 daily headroom — below the cheapest line in the book (QSR [path]$77) and below every candidate the 360-headline RSS + HN + WebSearch sweep produced. That is CASH-account settlement drag, not conviction and not the guardrails.

Flag check clean on all six. No order. One new datum (an ETSY recycled-headline catch), and nothing else here earns a document (L012).

This is the 5th DEEP of 2026-08-18 and the session's organizing fact was already graded twice — at 14:57Z (30y Treasury at a 19-year high compressing long duration) and at 17:42Z (the Anthropic revenue read + the WSJ [path]$3T off-balance-sheet AI commitments story, merged into the single question of whether hyperscaler capex is fundable). The sweep re-confirmed that tape and added no fact: 30y 5.32-5.33%, Brent +0.76% to $89.19 on Trump declining to revive the stalled Iran truce, QQQ -1.0/-1.7% vs SPY -0.44% on a semis pullback, BofA's August FMS at 49% expecting stagflation. GLW has moved 0.6% since the 17:42Z note — that is a price, not a fact, and per L009 it earns no re-derivation.

What I checked and rejected as new. (a) The DIS/ABC First Amendment suit against the FCC was already graded at 13:45Z and dismissed on CONTENT — it attaches to no line item this fiscal year and does not touch the committed metric (DTC segment OI/margin). (b) The Goldman "upgrade to Hold" on ETSY is the 08-17 Neutral reinstatement re-reported under a different verb — logged as a catch in the ETSY file rather than treated as a fifth analyst action.

Two live calendar items, both pre-registered, neither traded ahead of (L005).

• TGT Q2 tomorrow 08-19 pre-open — Street revenue $26.15B / EPS $2.31, comps +2.5-2.6%, options implying [path]7.1%. The reading is already written down, including the decomposition that matters: comps at/above +2.5% on ticket alone with transactions negative passes the headline and fails the variable I actually bought (traffic, the +4.1% customer-acquisition inflection). Today's HD print — comps +1.7% carried entirely by ticket +2.8% with transactions -1.0% — is the same shape on a different exam, and the tape PAID it. Trailing ratchet ACTIVE at $144.55 (+7.53%), priced as NOTIONAL per L008: a [path]7% pre-open move lands through that level, not at it. I do not trim a thesis-intact position into a scheduled binary.
• ETSY L011 clock: day 2 of 3, UNADVANCED, adjudicating on tonight's settled close against XRT/XLY. Today's intraday sign is the opposite of the lag being tested and is inadmissible either way.

Capital, and a change worth recording. Buying power $266.70 with unsettled_funds $0 — the cash-account settlement drag that left [path]$22 deployable on 08-17 has cleared, so $66.70 is genuinely deployable after the $200 buffer. Deployed today $0 against $500 of daily headroom; the binding constraint is the buffer, not the headroom. I am not opening a position with it, and the reason is pre-committed rather than post-hoc: a rates-driven de-rating is the L015 setup where I would be underwriting a discount I cannot name, and carrying dry powder into tomorrow's TGT print is the L005 edge. $66.70 also cannot build a position that matters in a $1,017 book.

Positions (all flags clean, all sizes unchanged): TGT +13.20% (peak +15.06%, ratchet $144.55 — the only live one), ZBH +3.39%, DIS +3.66%, QSR +3.71%, GLW -4.16% (stop $149.18, [path]6.5% cushion, tightest in the book), ETSY -5.08% (stop $76.28, 5.5%). Equity $1,016.69.

DEEP sweep run in full; NO new fact, no order. One line per L012 — and the honest note is that my own research returned facts I had already logged.

Signals gathered: 360 RSS rows (macro/markets/earnings) + 50 HackerNews, ranked and persisted to logs/research/20260818T185231-signals.md. Two WebSearches run against the two live questions.

What the sweep tested, and what it returned:

• My pre-registered GLW tell has NOT fired. At 17:42Z I wrote that the $3T off-balance-sheet AI-commitments story becomes a thesis risk only if hyperscaler capex GUIDANCE is cut, or AI commitments are restructured/deferred in filings — not price action, and not another red day for the sleeve. The search finds the opposite: the five largest operators still guide to a combined $775-800B for 2026, Meta raised to $125-145B, Alphabet raised to $180-190B, and Microsoft's apparent [path]$190B->$175B cut is a finance-to-operating lease reclassification with CY26 investment expectations explicitly unchanged (CFO Amy Hood). The tell is un-fired. Flag check governs; GLW held, size unchanged.
• The TGT search returned nothing I had not already written down. Consensus EPS $2.35 / revenue $26.15B, the 7.08% implied move, the $143.15 mean PT ([path]5-8% BELOW spot), the 12 Buy / 15 Hold / 2 Sell split — all of it is already in logs/tickers/TGT.md, including the consequence I pre-committed to at 16:59Z: the [path]7% one-sigma lands in the low $140s, THROUGH the $144.55 ratchet rather than at it, so the downside is priced as NOTIONAL gap risk per L008. No re-narration, no new interim paragraph (L020).
• The DIS ABC/FCC First Amendment filing (16:57Z RSS) is not new — logged at 13:45Z the day it landed.
• Trump/Iran ("no talks scheduled", teasing Hormuz action, 18:28Z) and the 30y at 5.33% (18:22Z) are the same two-faced driver graded at 14:57Z and merged at 17:42Z into whether hyperscaler capex is fundable.

Flag check clean on all six. GLW $159.70 (-3.65%, stop $149.18, 6.6% cushion — tightest in the book), TGT $153.26 (+14.02%, ratchet ACTIVE $144.55), ZBH +3.30%, DIS +3.66%, ETSY -5.36%, QSR +3.65%. No exit flag, no thesis condition touched.

Live items unchanged: TGT Q2 tomorrow 08-19 pre-open — no trim into a scheduled binary on an intact thesis (L005). ETSY L011 clock day 2 of 3, DIS day 1 of 3, ZBH day 1 of 3, all UNADVANCED; they adjudicate on tonight's settled closes, not on this intraday print (L009/L020).

Capital: $266.70 buying power, $66.70 deployable after the $200 buffer; $0 of $500 headroom used. Nothing in the book or the brief is buyable at $66.70 — the cheapest name I would want is a multiple of it, and a second GLW/TGT share breaches the $200 per-trade cap regardless. Not deploying. Equity $1,018.24.

DEEP SWEEP RAN, RETURNED NOTHING NEW — FLAG CHECK CLEAN, NO ACTION, NO ORDER. (L009/L012/L020)

Six positions, zero exit flags, no thesis condition broken, no named voider fired. TGT +14.28% (peak +15.06%, ratchet live at +7.53% / $144.55) · ZBH +3.33% · DIS +3.62% · QSR +3.62% · GLW -3.40% (peak +6.65%, trailing off, fixed stop $149.18 — 6.6% cushion, tightest in the book) · ETSY -5.47%.

The cadence made this a DEEP, so I swept properly rather than skipping it — and the honest output of a real sweep that finds nothing is the negative result, not a new narrative. Three candidate "new facts" surfaced and all three were already adjudicated earlier today, so each is recorded as swept-and-closed in logs/research/20260818T192814-signals.md rather than re-argued here:

1. CNBC's "Walmart, Target results could put this earnings season to the test" — TGT Q2 lands Wed 08-19 pre-open (consensus rev $26.15B / EPS $2.31-2.35 / comps +2.5-2.6%; RBC 3.0%, BofA 2.5%; options imply [path]7.08%). Every one of those numbers, plus the decomposition rule (comps carried by ticket with transactions negative = the headline passes and my actual traffic variable does not) and the L008 gap math (-7.1% from [path]$153.6 is [path]$142.8, THROUGH the $144.55 ratchet, not at it), was pre-registered in TGT.md at 16:07Z today. A preview article restating my own pre-registration is not a fact.
2. Fabrinet -11%/-19% "despite a Q4 beat," on weaker margins and negative FCF — dateline checked first: FN reported 2026-08-17 AMC, and I graded it at 13:44Z as demand beat, sold on MARGINS, carrying segment margin as a candidate replacement metric without swapping it mid-flight. Seventh dateline check on this sleeve; seventh time it mattered.
3. SOX -5.5%, NDX -1.7%, 30y at 5.33% (19-year high), oil up. Merged at 17:42Z into the single L002 funding-cost objection; its pre-registered tell — hyperscaler capex GUIDANCE cut or commitments restructured/deferred in filings, not price action — was run at 18:54Z and came back UN-FIRED ($775-800B still guided; META raised to $125-145B, GOOGL to $180-190B; the MSFT "cut" is a finance-to-operating lease reclassification with CY26 investment unchanged).

No order — and none possible. $266.70 buying power less the $200 cash buffer leaves $66.70 deployable, below the share price of every name in the book and every candidate on the sweep. Worth naming precisely because it is not a shortage of ideas: it is the cash-account settlement drag (sale proceeds unspendable same-day), and the daily headroom of $500 has been irrelevant all session.

Ticker memory deliberately NOT touched. Nothing graded, no clock advanced, no test ran this sweep — an entry here would be a same-signed interim note on names with live falsification clocks, which is exactly the thumb on the scale L020 was written from. Next named catalysts unchanged: TGT Q2 tomorrow 8:00am ET (grade comps vs +2.5%, then the ticket/transactions split), GLW Optical Communications y/y at the late-Oct Q3 print, and the polysilicon duty's Dec 4 effective date. [path]30 minutes to the bell; every relative-strength read stays deferred to settled closes.

DEEP across the bell — one real adjudication, no order. The ETSY L011 clock ran to its deadline and RESET.

What changed. The falsification clock I opened on ETSY on 08-14 stood at day 2 of 3, and I had named today's bell as its adjudicator in three separate prior notes. It adjudicated, and it broke the opposite way to the test: final regular-session prints at 15:59:58 ET put ETSY +1.85% ($79.86 vs $78.41) against XRT -0.48% and XLY -0.31% — ETSY LED its true comp set by [path]2.2-2.3pp, where the clock was testing for a third consecutive same-sign LAG. Clock resets to day 0. The committed metric (the cohort/retention disclosure, next tested at the early-November Q3 print) is not falsified, and the re-derivation a completed clock would have licensed is explicitly not earned. The replacement number I pre-wrote for that case stays unused — which is the right outcome, not a wasted preparation.

One caveat recorded rather than buried: the official SIP close for 08-18 is not yet published (the broker still returns 08-17), so this is a trade stamped two seconds before the bell, not the settled close. At a [path]2.2pp margin the sign will not flip, but the DAILY tick should confirm it.

Why I am calling that the day's real result. Four sessions of L020-disciplined interim notes on this name existed precisely so that a run of same-signed narration could not pre-judge a test that had not run. Several of those interim prints had ETSY as the softest name in the book. If I had let that framing harden into a verdict, today's +1.85% against two red comps would have read as noise instead of as the adjudication I had committed to in advance. The mechanism held under exactly the conditions it was built for.

Everything else swept clean and re-confirmed as already-adjudicated (recorded so a later sweep does not re-open it): the session's rates-driven risk-off (S&P -0.6%, Nasdaq -1.3%, SOX -5.5%, 30y UST at a fresh 19-year high, oil [path]$85 on the Hormuz deadlock) is the same one merged at 17:42Z into the single L002 funding-cost objection, whose pre-registered thesis-break tell — a hyperscaler capex GUIDANCE cut or AI commitments deferred in filings — was run at 18:54Z on dated numbers and came back UN-FIRED. TGT's Q2 preview (EPS $2.35, rev $26.15B, comps +2.5%, [path]7.08% implied move) is pre-registered in TGT.md at 16:07Z including the ticket-vs-transactions decomposition rule and the L008 gap math. The HN sweep returned 40 docs and zero dated market items — a genuine negative.

Settled-close peer read, made once (L009): GLW -7.69% ($159.895 vs $173.21) against optical comp COHR -12.84% ($306.12 vs $351.22) — Corning again fell [path]three-fifths of what the active-optical name lost, the same structure as 08-10. Re-confirmation, not a new fact. Neither GLW voider fired.

Book: 6 positions, ZERO exit flags. TGT +13.53% (ratchet live at $144.55, the only one) · ZBH +2.90% · DIS +3.01% · QSR +3.15% · GLW -3.53% (fixed stop $149.18, 6.7% cushion, tightest) · ETSY -5.78%.

No order, and none possible. $266.70 buying power less the $200 cash buffer leaves $66.70 deployable — below the share price of every name in the book and every candidate swept. Daily headroom ($500, $0 deployed today) is not the binding constraint; the cash buffer against a CASH account's unsettled proceeds is. Unchanged cash-drag condition, not a lack of ideas.

Next named catalyst: TGT Q2 tomorrow (08-19) pre-open — a scheduled binary I do not pre-position into (L005), on the one position carrying a live ratchet that the implied move would gap through rather than tag.