DEEP — a fired trigger declined, and $234.86 earmarked to a date

Tape. Trump called off the "massive" strike on Iran and talks resume today; oil slid, the Dow ran +600 and the Nasdaq +1%. Every one of my five positions is green. None of that is mine (L006) — TGT +3.7%, ZBH +2.6%, CMG +1.5%, KMX +1.4%, XYZ +0.6% is what a book of long equities does when a geopolitical discount comes out of the market at 9:30. I searched TGT specifically for an idiosyncratic reason behind its +3.7% and found a cookbook press release and a CNBC "Final Trades" mention. That is not a reason. Recorded once, not re-argued.

Exits: none. No flag lit. KMX carries the tightest cushion (spot $58.04 against a live ratchet at $56.92, [path]1.9%), TGT's ratchet stepped up to $142.11 on a new high-water mark, CMG/XYZ/ZBH are still on their fixed -10%. Every break-line I pre-committed to is untouched: no Chipotle-inside-the-outbreak disclosure, no CarMax guidance or CAF credit crack, no Block GP stall. Per L009 the job on an unflagged position is to check and stop, so I checked and stopped.

The one real decision: GLW's re-arm trigger fired and I declined it. Truist upgraded Corning to Buy from Hold today, PT $175 — cut from $205 — with spot at $138.07, [path]27% below the target. My 06-08 memory said I would re-arm on "a genuine fresh sell-side UPGRADE (rating change, not just PT bumps) with price well below target," and this is exactly that, on the cheapest valuation gate I have seen all month. I am still flat, because the clause that actually matters went unmet. My 07-28 note — written the day GLW fell 15.5% on optical-segment deceleration — set the bar as the AI-capex bellwether guiding capex back UP, and Truist's note is a valuation call: a better entry point after a [path]45% drawdown. A firm that cuts its own target 15% while raising the rating is saying the cycle got worse and the price got cheaper faster. That is the de-rating I sidestepped wearing a Buy sticker.

The tape refuted it independently and I only needed to look once: GLW is -0.13% on a +1% Nasdaq day. An upgraded name that cannot get green on a broad relief rally is the market declining the upgrade.

I am recording the counter-argument so this is gradeable rather than self-flattering: Q2 Optical grew 32% to $2.07B with Enterprise Networks +65%, and the NVIDIA deal expands US optical connectivity capacity 10x. If GLW runs from here, that is why. I weight it below the deceleration read because capacity expansion is a promise about demand while sequential decel is a measurement of it — and a -45% July says which one is being priced.

The rest of today's sheet, screened and passed: IR (Stifel Buy, PT $101) is a +4% chase only [path]14% under the new target — no gate. LYB (JPM Overweight, PT $80) is red on its own upgrade in a commodity-chemical downcycle I have no edge in. ETN's $412 share breaks the $200 per-trade cap. RIVN is the pattern I refuse on principle — a Q2 beat de-rated -9.6% because the load-bearing metric (auto margin, cash burn) didn't move. Note what all four of these have in common with my entire book: they are the same fresh-buy-rated-PT template. Running it a sixth time this morning would be L004 exactly as written — one bet that the template works, dressed as a new conviction.

Capital: $234.86 over the buffer is RESERVED, not idle, and it has a date on it. Two of my own positions print Wednesday 2026-08-05: ZBH Q2 before the open, where the pre-named wall is organic growth vs the Street's [path]3% ceiling, not EPS; and XYZ Q2, where the wall is gross-profit growth, the exact number XYZ's add-gate has been waiting on since 07-27. Payrolls and ISM land in the same week, with a divided Fed that just had three members vote to hike. L005 is unambiguous about this shape: carry dry powder INTO a dated binary. If ZBH prints mid-single-digit organic, that is the second independent signal that has been missing and I want cash to act; if it prints ≤3%, the thesis voids and I sell. Spending that cash today, on a valuation upgrade, during a relief rally, 48 hours before the two numbers I have already committed to being judged on, is the worst available use of it.

Orders placed: none. Next DEEP acts on the 08-05 prints — on a fact, not on a new price.

DEEP — swept, no new fact, no orders. RSS clean (360 docs) + 410 ticker-news rows; HN stale; WebSearch degraded (2 of 3 queries unavailable, the third served 2023 results) — logged in the brief rather than passed off as a full sweep. Only fresh wire item since the 13:44Z DEEP is the confirmed U.S.-Japan yen intervention: macro, no read-through to any of the five. Flags clean (KMX $58.35 vs ratchet $56.92; TGT $148.51 vs $142.11; CMG/XYZ/ZBH on the fixed -10%). $234.86 over the buffer stays RESERVED into Wed 08-05: ZBH Q2 (organic growth vs [path]3%) and XYZ Q2 (gross-profit growth). Next DEEP acts on those prints, not on a new price.

DEEP — full sweep run, no orders. This is a DEEP that changed the evidence, not the position.

What was new: the sweep itself worked. The 14:20Z DEEP had to log itself as degraded — two of three WebSearch queries returned "unavailable", the third served 2023 results — so its no-trade call rested on a sweep I couldn't stand behind. This one ran complete: 360 RSS docs (macro/markets/earnings), 387 per-symbol news rows, and all three WebSearch queries answering with live 2026-08-03 results. The conclusion is the same but it is now load-bearing: I looked properly and there is nothing here, rather than being unable to look.

What the working sweep surfaced, and why none of it is a trade. The X/fintwit tape is the crowded megacap-AI complex (NVDA, MSFT, MU, PLTR, AAPL, AMZN, TSLA, INTC) — no edge, no catalyst, no size that matters at $200/clip. The WSB leader is ENvue Medical on a +46% 24h mention surge, which is a microcap mention-spike and exactly the shape I don't touch. Macro since 13:44Z: confirmed U.S./Japan yen intervention (13:48Z), Trump calling off the Iran strikes with talks resuming today (14:38Z), oil -6%. None of that reads through to a burrito chain, a used-car retailer, a big-box retailer, a payments processor or a knee-implant maker. The only CMG-specific wire item is dated 2026-07-30 and is the 07-29 Q2 print already in memory — a restatement, not a signal (L004).

One near-miss worth logging, because I caught it in the right order. The sweep handed back UBS's buy / $115 PT on ZBH and I started to write it up as corroboration two days ahead of the print. Reading ZBH.md first — as the discipline requires — showed that call is the entry thesis, in the file since I opened the position. A search engine returning my own thesis is the purest L004 trap: one signal, re-encountered, dressed as a second. It earns no size. Filed as a near-miss, not an insight.

L006 applied to myself before I enjoy the day. All five positions are green — TGT +2.9%, ZBH +2.1%, CMG +1.8%, KMX +1.4%, XYZ +0.5% — on a tape explicitly described as a catch-up trade where small-caps and blue-chip industrials lead and tech lags. I am long five value/cyclical/consumer names on the day that style box leads. The L006 test (is any name uniquely green while its peers fall?) returns no for all five. This is factor beta, and ZBH.md already recorded it once at 13:45Z, so it is not being re-recorded per name (L012).

Book: no exit flags. KMX +9.94% vs ratchet $56.92; TGT +10.64% vs ratchet $142.11; XYZ +6.54%, CMG -0.61%, ZBH -0.62% still on the fixed -10%. Nothing within reach of a stop. No ticker memory appended this tick — no symbol produced a new fact, and appending a restatement is the exact failure L012 was written from.

CAPITAL: $234.86 over the $200 buffer stays RESERVED. ZBH Q2 and XYZ Q2 both print Wed 2026-08-05 — two of my five positions. For ZBH the pre-committed clean number is ORGANIC growth vs the Street's [path]3% (which is the ceiling of management's maintained 1-3% FY band), not headline EPS (L010). Deploying today would mean buying day two of a geopolitical relief rally, into a rotation I've just admitted I'm riding for free, 48 hours before two of my own load-bearing prints. L005 is written for precisely this: carry the powder INTO the binary. Next DEEP acts on the 08-05 prints.

DEEP — full sweep, no trade, reserve intact.

Sweep: RSS 360 docs (macro/markets/earnings) clean, tick per-symbol capture 392 rows, HN 20 docs contributing nothing, WebSearch live on the news query and stale/aggregator junk on the two social ones. Nothing on the wire is new for this book — the only position-adjacent items in 360 documents are both dated 07-30 and already sitting in CMG's memory (the comp-guide raise; Yum's "Taco Bell sales already improving"). Today's real movers (GOOGL +3.4% on a mark-to-market windfall, AZN -6% on the BMY approach, TGTX -9.8%) are in sleeves I do not own and have no edge in. The tape is still Friday's story running on: Iran talks resume, oil slides, small-caps and industrials lead, tech lags.

The one piece of actual work this tick: ran the L011 falsification test on CMG — the only red name in a green book — once, on settled closes, against the QSR comp set rather than the index. It does NOT fire. 07-30: CMG +12.5% vs YUM +3.3% / CAVA +1.7% / MCD -1.1% / WING -3.7% (idiosyncratic upside on its own print). 07-31: CMG -3.4% vs YUM -2.4% / WING -3.6% / CAVA -1.8% (mid-pack in a complex-wide give-back — sector beta, not a message). One ambiguous session, not the three L011 requires. Tripwire logged to CMG.md so the next tick inherits it instead of re-deriving: two more sub-complex closes (Tue + Wed) makes three and earns ONE re-derivation.

No exit flags. KMX +10.16% (ratchet $56.92), TGT +10.33% (ratchet $142.11), XYZ +6.65%, CMG -2.69%, ZBH -0.61%. Nothing near a stop.

CAPITAL: $234.86 over the buffer stays RESERVED. ZBH and XYZ both print Wed 08-05 — 40% of the book through a binary in 48 hours — and today's ZBH preview sharpens the pre-named clean number (organic growth; Street wants $2.13B / +2.5% revenue and $2.01 EPS / -2.9% y/y, with sluggish organic growth and robotic-surgery competition named as the previewers' bear case too). L005: carry the powder INTO the event, don't spend it the day before. Next DEEP acts on the Wednesday prints.

DEEP — full sweep run, no new fact, no order. Reserve unchanged.

RSS 360 / HN 30 / per-symbol capture 473 / WebSearch across QUERIES (news live, social queries degraded to aggregator pages). Four position-adjacent hits, all restatements already in ticker memory — the two 07-29/07-30 CMG items, and a Carvana guidance headline that is the 07-29 CVNA guide-below-Street print already owned in the 13:45Z KMX note. Nothing earned re-derivation, so nothing got it (L009/L012).

No exit flags. KMX +9.55% vs ratchet $56.92, TGT +10.94% vs ratchet $142.11, XYZ +6.43%, CMG -2.29%, ZBH -0.60%.

$234.86 above the buffer stays reserved into Wednesday: ZBH pre-open (organic growth vs Street [path]3%) and XYZ post-close (gross-profit growth + Cash App actives). L005 — carry the powder in.

CMG L011 tripwire inherited, not re-derived: needs a sub-QSR-complex close on BOTH Tue and Wed; one session exists. Next DEEP acts on the Wednesday prints.

DEEP — no order. One new fact, and it is a correction to my own record.

TGT: Wolfe Research raised its PT to $169 from $162 (Outperform maintained), dated today. The 13:45Z DEEP note stated there was no TGT-specific catalyst and booked the day's +3.7% entirely as Iran-relief beta. That was wrong — the Wolfe item published 10:26Z and was already in the news log at the 13:40Z capture, before the note that missed it. TGT is +3.31% into midday while the Dow runs [path]+1%: it is leading its tape, and there is now a name-specific reason why. The failure mode is worth naming because it is L006 run backward — I was careful not to credit a rotation day as skill, and that care became the reason I stopped screening the symbol's own wire. Explaining a move with the tape is not a substitute for checking the name.

The substance: Wolfe named the exact metric I pre-committed to (L010). The clean number for the 08-19 print, written down six sessions ahead, is comps + traffic + gross margin against markdown/shrink — explicitly not headline EPS. Wolfe now models Q2 comps 3.0%, Q2 gross margin above expectations, and low-to-mid-single-digit comps in H2 toward [path]$10 of 2027 EPS at 17x. The bar is externally specified now: a 3-handle comp with clean margin confirms; a 1-2% comp — or comps bought with markdown-driven traffic and margin compression — breaks the thesis whatever EPS does.

Not an add trigger. Third Wolfe action on the name, second PT raise, same analyst the position was opened on. The 07-30 ticker note already ruled on this shape: a reiterate-with-a-PT-raise is one bet restated, not a second independent signal (L004). Add-gate shut, TGT stays a starter.

Capital unchanged and not re-argued (L005): $234.86 above the $200 buffer stays reserved into Wednesday — ZBH pre-open (clean number: organic growth vs the Street's [path]3%) and XYZ post-close (gross-profit growth + Cash App actives), [path]23% of book through two binaries in 48 hours. Carry the powder in, don't spend it the day before.

Macro, credited once: July ISM manufacturing 55.6 vs 54.0 consensus, up from 53.3 — a real expansion beat, supportive of the consumer/cyclical sleeve, not a thesis fact for any name.

Flags: clean, none close. KMX +9.26% vs its +7.73% ratchet ($56.92), still the tightest cushion in the book. TGT +11.05% with the ratchet now live at +5.72% ($142.11). XYZ +6.49%, CMG -2.47%, ZBH -0.78% — all on the fixed -10%. Other position-adjacent hits were restatements and got one line each in the brief (L012).

DEEP — full sweep run, no new fact, no order. RSS 360 + HN 30 + per-symbol capture 406 rows + WebSearch: zero ticker signals, nothing on a held name. The only position-adjacent hit ("Why TGT Stock Is Trading Up Today", 16:45Z) chased down to the same Wolfe $162->$169 Outperform raise already owned at 16:35Z — second wrapper on one catalyst, so the add-gate stays shut (L004), and TGT's fresh 52-week high is a price, not a fact (L009). Flags clean on all five: KMX +9.21% vs its +7.73% ratchet ($56.92, tightest cushion), TGT +11.01% vs +5.72% ($142.11), XYZ +6.65%, CMG -2.49%, ZBH -0.46% all on the fixed -10%. The $234.86 above the buffer stays reserved into Wednesday's two binaries — ZBH pre-open and XYZ post-close on 08-05, [path]23% of book in 48 hours (L005). Next DEEP acts on those prints.

DEEP — full sweep, no order. Two new facts, so this gets a real note rather than the one-liner L012 mandates for an unchanged tick.

1. The TGT add-gate reason I have been re-using is now dead, and I am saying so instead of recycling it. Gordon Haskett reiterated Buy at a $170 PT; Jefferies is at $161 and Wells Fargo Overweight at $165. Every prior note refused the add because Wolfe raising Wolfe's own target is one bet restated, not a second source (L004). Gordon Haskett is a different shop — L004's second-signal test is met, with a 52-week-high breakout alongside it.

The add is blocked by the per-position capital cap, and the validator would NOT have caught it. config.yaml reads per_trade_max_usd: 200 = "max capital in any single position"; TGT already holds $149.58 of it, so one more share ([path]$299 total) is [path]50% over. But guardrails.evaluate only tests the ORDER's notional — $149.58 < $200 passes, the daily cap passes ($0 deployed), the kill-switch passes ($1,026.78), the cash buffer passes ($285.28 left > $200). A 1-share TGT buy would validate APPROVED today. I am not taking the trade the check would have allowed, because the cap the config states is per-position and the cap it enforces is per-order. Flagging that mismatch for the human as the substantive finding of this tick — every position in this book is currently under $200, so the gap has never bound before; TGT is the first name large enough to expose it.

2. XYZ's clean number now has figures, two days early. Company guide Q2 gross profit $3.04B / +20% y/y against a Street EPS $0.48 / -22.6% y/y, with Cash App MTA consensus 59.17M (vs 57.00M) and preliminary Cash App Bitcoin ecosystem revenue $1.8B. That divergence is textbook L010: the headline line is guided to fall while the line the thesis owns is guided to grow 20%. Verdict pre-committed both ways — gross profit vs the $3.04B guide and MTA [path]59M decide it; an EPS beat does not rescue a gross-profit miss, and the guided EPS decline is not a break.

ZBH: FY26 organic CC guide confirmed at 1-3%, so [path]3% is the ceiling of management's own guidance and a Q2 organic print at/below 3% voids the thesis. Wall unchanged, now sourced.

CAPITAL — the $234.86 above the buffer stays reserved, reasoning unchanged (L005): ZBH pre-open Wed 08-05, XYZ post-close Wed 08-05, [path]23% of book through two binaries inside 48 hours. Sweep found nothing to buy anyway — zero ticker signals from 360 RSS docs, and the live Stocktwits trending list (GME, MLTX, RUM, BA, TGTX, STLA, CELH, SOFI, PLTR) contains no name I have an edge on.

Book: no exit flags, none close. KMX +9.53% vs its +7.73% ratchet ($56.92) — tightest cushion. TGT +11.28% vs +5.72% ($142.11). XYZ +6.67% and CMG -2.41% and ZBH -0.36% all still on the fixed -10%. Today's green across all five is the Iran-talks relief tape (oil -6%, QQQ +1.4%, ISM 55.6), credited once as beta, not alpha (L006).

DEEP — full sweep run, zero signals, no new fact, no order. Stance from the 17:39Z DEEP carries unchanged.

• Sweep: RSS 360 docs (macro/markets/earnings) + HN 30 + 400 per-symbol news rows captured on the tick + WebSearch across the news/stocktwits QUERIES. 0 ticker signals extracted. Brief persisted: logs/research/20260803T182148-signals.md.
• Exit flags: none, none close. KMX +9.67% vs its +7.73% ratchet ($56.92); TGT +11.51% vs +5.77% ($142.17); XYZ +6.82%, CMG -2.56%, ZBH -0.08% all still on the fixed -10%.
• Checked each held name for a new fact; found none. CMG's searchable record is still the 07-29 print; KMX's is still the 07-21 Barclays / 07-29 JPMorgan upgrades (next print 09-24); TGT/XYZ/ZBH nothing past the 17:39Z brief. Per L009 a fresh price is not a fact, so nothing was re-derived.
• No deployment. Two standing reasons, neither new: (1) L005 — ZBH pre-open Wed 08-05 and XYZ post-close Wed 08-05 put [path]23% of the book through two binaries inside 48 hours, so the $234.86 above the $200 buffer is carried in, not spent the day before; (2) the TGT add stays blocked by the per-position cap ($149.58 of $200; +1 share -> [path]$299, [path]50% over) — the validator would pass it because it tests the order's notional, not the resulting position. Already flagged for the human at 17:45Z; restated once so it survives between briefs, not re-argued.
• Ticker memory: deliberately not appended this tick. The sweep produced zero new observations on the five held names; writing "nothing new" into five files an hour after the last update is the exact L012 failure mode (a restatement acquiring unearned confidence by being freshly typed). The pre-committed walls are unchanged and already on record — ZBH: organic constant-currency growth vs the 1-3% FY guide, not headline EPS (L010); XYZ: gross profit +20% y/y vs the $3.04B guide and 59M+ Cash App MTA, with the guided -22.6% EPS decline expected rather than the verdict.
• Tape credited once (L006): Iran-talks relief rally — oil -6%, dollar index at a seven-week low, ISM manufacturing 55.6, Honeywell +7.5% leading the Dow. Today's green marks in this book are that beta, not alpha. 136 S&P names report this week; July NFP Friday.
• Next: the Wednesday prints are the next thing that earns an opinion. Until then, check the flag and stop (L009) and keep the output to one line (L012).

DEEP — a real new fact landed on CMG, so this tick earns a full note (L012). No order placed.

What changed. Michigan confirmed the first two US deaths in the cyclospora outbreak today — >11,000 cases in that state alone, 34+ states affected, and the CDC now expects illnesses to keep INCREASING through the end of August. That kills one leg of my own 07-30 CMG entry case, which argued the onset curve (06-22 -> 07-20) pre-dated the 07-17 Taylor Farms recall and "should roll off." It isn't rolling off. I'd rather record that against myself than let it sit in the note as though it were still true.

Why I did not sell. The 07-30 pre-commitment was deliberately narrow and made once: exit on a HEADLINE putting Chipotle inside the outbreak. Today's traceback runs the other direction — the nine-state event is Taylor Farms de Mexico iceberg lettuce served at Taco Bell, with further clusters at Olive Garden and Red Lobster; CMG's lettuce is California-sourced and the company states it is "not involved." The separate 72-person outbreak still has no identified source, which is the live tail risk and remains the trigger. Competitors implicated, CMG not. Pre-commitment unfired, and a worse tape is not a headline (L009). Held.

The number I should have written down at entry, now named (L010). CMG guided Q3 comps to [path]+1%, explicitly including a [path]200bp cyclospora / consumer-caution drag — the FY low-single-digit comp guide the thesis rests on sits on top of that. That is the bar the Q3 print gets graded against, not headline EPS. A longer outbreak season is precisely what makes 200bp too light, so the new fact and the graded metric are the same variable. Written into CMG's memory so the next print is judged against a number I named in advance rather than one I reverse-engineer after it.

L011 test, run once, on settled closes. 07-31 is lag session 1 of 3: CMG -3.37% vs XRT -0.88%, MCD +0.82%, QSR -0.43%, DRI -1.64%, YUM -2.37% — CMG behind every comp. Today is shaping as a second (CMG -0.11% against a +1.5% S&P / +2.0% Nasdaq — the only name in the book not joining the Iran-relief rally) but is unsettled and does not count; the entire point of L011 is that the read happens on settled closes. At 3+ same-sign settled lags I re-derive which number the market is actually pricing, once.

Tape credited once (L006). Trump called off the Iran strike to negotiate the Strait of Hormuz — oil -6%, Nasdaq +2.0%, S&P +1.5%, Dow +1.1%, Amazon through $3T. Every green line in this book today is that beta, not stock-picking. Which is also the tell that CMG's flatness is idiosyncratic.

No deployment, unchanged grounds. (1) L005 — ZBH prints pre-open Wed 08-05, XYZ post-close Wed 08-05; [path]23% of the book through two binaries inside 48 hours, so the $234.86 above the $200 cash buffer is carried INTO them, not spent two days early. (2) The TGT add is still blocked by the per-position cap ($149.53 of a $200 ceiling); the guardrail gap raised for the human at 17:45Z — guardrails.evaluate tests the ORDER notional, not the RESULTING position — is not re-argued here.

Book: no exit flags, none close. KMX +9.99% vs its +7.73% ratchet ($56.92); TGT +11.24% vs +5.77% ($142.17); XYZ +6.90%, CMG -2.49% (stop [path]$34.32), ZBH -0.15%, all on the fixed -10%. Sweep itself was empty for the third straight DEEP: 390 docs, zero ticker signals; Stocktwits trending (USO, PLTR, CRML, BMY, TGTX, AZN, IONQ, UCO, CRCL, DECK) held nothing I have an edge on. Next DEEP acts on the Wednesday prints.

DEEP — full sweep run, no new fact, no order. Book unchanged.

Sweep: RSS macro/markets/earnings + HN (390 docs), 408 ticker-news rows captured, WebSearch across the news/stocktwits QUERIES. Zero ticker signals extracted — fourth consecutive DEEP with that result, same cause (CNBC trader-talk/buffett-watch blocks serving evergreen copy; the only dated items are Iran-relief market wraps). Social trending: TMTG, JOYY, Strive, Sprout Social, Weibo, IONQ/RGTI/QBTS/CRWV — no held name, no edge. Discarded.

Checked every held name directly: nothing new since the 18:55Z DEEP. CMG returns only the 07-29 print already in memory (EPS $0.33 adj vs $0.32, rev $3.35B, comps +2.2% on +1% traffic, FY comp guide raised to low-single-digit, [path]2pp cyclospora hit to late-July sales). ZBH returns only the 08-05 pre-open print and the 07-28 UBS upgrade. No headline places Chipotle inside the outbreak, so the 07-30 pre-commitment stays unfired.

No exit flags, none close. KMX +10.30% vs its +7.73% ratchet ($56.92), TGT +11.23% vs +5.77% ($142.17), XYZ +7.16%, ZBH +0.03%, CMG -2.08% (fixed stop [path]$34.32). L011 lag count on CMG still 1 of 3 — today is shaping as a second but is unsettled and does not count; the 08-04/08-05 tripwire in the ticker memory stands as written.

No deployment (L005): ZBH prints pre-open Wed 08-05 and XYZ post-close Wed 08-05 — [path]23% of the book crosses two binaries inside 48 hours, so the $234.86 above the $200 cash buffer rides INTO them rather than being spent two days early. TGT add still blocked by the per-position cap ($149.51 of $200).

Next named catalyst: the Wednesday prints (ZBH pre-open, XYZ post-close, 08-05).

DEEP — full sweep run, no order. One real new fact today: the settled close.

CMG closed GREEN +0.66% ($37.465), reversing the intraday read. The 19:28Z brief had today "shaping as a second L011 lag session" off a -2.1% intraday print, correctly flagged UNSETTLED. It rallied into the bell. Lag count stays 1 of 3 — and on settled closes against the TRUE comp set CMG LED rather than lagged: +0.66% vs MCD -1.99%, QSR -1.49%, YUM -2.88%, DRI +1.34%, three of four restaurant peers red on a +1.42% SPY / +1.94% XRT day. That is L006-idiosyncratic (uniquely green while the sector falls), not style-box beta. It trailed XRT by 1.28pp, but XRT is broad-retail beta on an Iran-relief tape, not Chipotle's comps. The pre-commitment to count only settled closes is the reason this did not become a decision — exactly the failure mode L009 was written from.

The peer split now pays the outbreak leg rather than just failing to break it. YUM — Taco Bell's parent, the implicated chain — was the worst restaurant close of the day at -2.88% while CMG held green. The market is pricing the outbreak into the implicated name and not into Chipotle.

CMG pre-commitment UNFIRED, checked directly. FDA's second, separate 72-case cluster still has no identified product, location or supplier (The Hill / CIDRAP today). Escalation is on the already-known outbreak — 45 states (from 34+), 2 deaths, CDC expecting cases to rise through end of August — traceback still Taylor Farms de Mexico lettuce at Taco Bell, CMG California-sourced and "not involved." No headline places Chipotle inside either outbreak, so no exit.

No deployment, unchanged grounds (L005). ZBH prints pre-open Wed 08-05, XYZ post-close Wed 08-05 — two binaries inside 48 hours across [path]17% of the book. The $234.86 above the $200 buffer is carried INTO them, not spent two days early. The bell has also rung (20:00Z), so nothing routes at market anyway. TGT add still blocked by the per-position cap ($149.31 of $200).

Book at the close: no exit flags, none close. KMX +10.49% vs its +7.73% ratchet ($56.92) — tightest cushion at [path]2.5%. TGT +11.07% vs +5.77% ($142.17). XYZ +7.10%, ZBH +0.40%, CMG -1.90%, all three on the fixed -10%. Sweep otherwise dry: social trending (CRWV, LKNCY, REPL, ZETA, RDDT, RBLX, COIN, LMND, neoclouds) holds nothing I own or have an edge on; HN evergreen. Counter-signal noted for later: ISM inflation worries "worse than pandemic era" and Kalshi pricing a cool July payrolls — the relief rally meets a live stagflation print Friday. Next DEEP acts on the Wednesday prints.