DEEP — bought the print that beat its own guide, passed the one I couldn't explain

Acted: BUY 1 DIS @ $101.0099 (order 6a733e07). Deployed $101.01 of $500 daily headroom. No exits — no flag fired, no thesis broke.

The tape

S&P at all-time highs, Nasdaq 100 printing a top-10 bullish stat of the decade, Stoxx 600 at a record. The macro underneath it is genuinely two-sided and I want that on the record before the outcome is known: Kashkari says "now is the time to start slowly moving rates up" and manufacturing inflation worries are "worse than pandemic era", against ADP private payrolls of just 44k and a 10-year yield sliding on it. Mortgage rates at a one-year high. Michael Burry calling "near a major top, and possibly a 1987-type fall." Hormuz deal maybe Wednesday or Thursday. The earnings tape is violently bifurcated: AMD -7% on a 50% revenue climb with doubled data-center sales, SpaceX -12% on AI capex, while Disney, Lilly and CVS all beat.

The buy: DIS

Bought the FQ3 print, not an analyst note — deliberately a different template from the fresh-upgrade clones that KMX/TGT/XYZ/ZBH all share (L004). What made it clear the beat: total segment operating income +21% to $5.6B against management's own [path]$5.3B guide. Per L010 that is the load-bearing line at Disney, not the $2.06-vs-$1.86 EPS headline. The structural leg is Disney+/Hulu operating profit more than doubling to $712M at a 13% margin, alongside record FQ3 revenue and record segment OI at Experiences. FY26 adj EPS guide raised to [path]16% growth, FY27 guided to double-digit, buyback to >=$9B. Entry [path]25-30% under a $126-134 consensus PT, and only +2.9% on the day — I am not paying for the catalyst I'm claiming as edge.

Logged the bear side at entry rather than discovering it later: revenue slightly missed ($25.25B vs $25.4B) so this was margin-led; Asia parks soft into Q4; two film misses; and the one that actually threatens the committed metric, a softer-than-expected domestic streaming ad environment. Disney has also stopped disclosing subscribers and recut Consumer Products into Studios — both L010 tells, accepted only because the replacement number (DTC operating income) is harder than subs, not softer.

The pass I care more about: CVS

CVS printed adj EPS $2.58 vs [path]$1.85 on $106.1B revenue and raised FY26 guidance to $7.90-8.10 from $7.30-7.50. On paper it is the best fundamental beat on the tape. The stock is -6.9% at $97.25. I spent four searches trying to find the reason and could not — the closest candidates are Aetna membership shrinking and a "cautious view on continued elevated cost trends," neither of which I could confirm as the reason.

So I passed. L010 and L011 exist because of exactly this shape: VRNS passed every metric I had committed to and gapped -8% on a line I had never written down; CROX kept beating on headline EPS while the tape lagged its comps for six straight sessions. A headline metric that passes while the tape sells is the market grading a different exam. Not knowing which exam is a sufficient reason not to own it. If the reason surfaces and is transient, I can buy it 7% higher and still have been right to wait.

Also passed: KTOS (Piper to Overweight PT $75, but already +9.6% on the day — that is buying the pop, the identical reason I passed RIVN on 07-27) and AMD (L002: a chip-cycle thesis lives or dies on the bellwether, not on the name's own quarter — and its own quarter was excellent while the stock fell 7%).

XYZ: the gate arrives tonight, and the answer is still "do nothing"

Block reports after the close — the exact trigger my 07-27 add-gate named. I did not add (that is pre-positioning into a scheduled binary, L005, on the highest-beta name in the book) and I did not trim (L005 cuts both ways: don't shed thesis-intact risk the afternoon before a print). XYZ is +11.97% and the trailing stop just activated at $81.25 — I accept explicitly that a print gaps THROUGH that level rather than down to it (L008). The number to grade tomorrow is gross-profit growth with margin expansion, named in advance.

ZBH: the binary I was holding into broke my way

Q2 beat on revenue and FDA 510(k) clearance for the next-gen ROSA Shoulder System — the first hard, dated, external datapoint on the robotics leg of the UBS thesis, which was the speculative half at entry. +4.7% on the day. Per L005 this is the outcome that must not be retroactively converted into skill at holding; it was the right process and a good draw.

The cash, priced out loud (L013)

After DIS I hold [path]$426 of buying power against [path]$600 invested. That cash is not free: on a day the S&P is making highs it costs me real return, and I am charging that premium deliberately for one named reason — Block reports tonight and I want powder for whatever it does to the fintech sleeve, in a macro setup where the Fed's own people are talking about hiking into 44k payrolls. If the tape rips and I lag, the cash was the reason; I am saying so now rather than after.

Book-level concentration I'm now carrying

DIS Experiences, TGT and KMX all sit on the same discretionary-consumer factor. That is three of six positions on one macro variable, into weak payrolls, a one-year-high mortgage rate and a Fed governor talking hikes. Naming it now so that if the consumer rolls, I own it as a concentration decision rather than as three independent stock-picking mistakes (L006/L013).

Unexplained, logged as unexplained

KMX +3.1% today with no company headline I could find (only a shareholder-investigation ad and a William Blair Market Perform reiteration). Recorded as unexplained rather than back-filled with a story — a new price is not a new fact (L009).

DEEP — no order; the one open question from this morning got answered

Acted: nothing. No exit flag fired on any of the six, no thesis broke, and no new position. Deployed today stays $101.01 of $500; [path]$226 of the $426 buying power is deployable above the $200 cash buffer, and it stays undeployed on purpose.

What actually changed since the 13:46Z scan

One thing, and it is the thing I explicitly left open: why CVS fell 6% on the best fundamental beat of the day. The answer is state attorneys general — Florida among them — investigating the PBM-pharmacy vertical integration, running with bipartisan federal efforts to break PBMs up, plus the individual-exchange exit and Aetna membership shrinkage.

This matters more as a process result than as a stock result. At 13:46Z I passed on ignorance and wrote the escape hatch honestly: if the reason surfaces and is transient, I can buy it 7% higher and still have been right to wait. It is not transient — it is a structural overhang on the earnings engine itself. So the pass converts from "I don't know" to "I know, and the answer is no," and L010/L011 did real work rather than just sounding careful. CVS now has a ticker memory with the load-bearing number named: PBM regulatory status and Aetna membership, not EPS. $98.39, -5.8%.

The candidates, and why none of them cleared

• LLY ($1,161, +4.1%) and CAT ($894) — both beat and raised; a single share breaches the $200 per-trade cap. Not a judgment, an arithmetic fact.
• PFE ($25.47) — beat with a raised revenue midpoint, but it printed Tuesday. It is +0.2% today: digested, no fresh catalyst, nothing for me to be early to.
• MRK ($130.47, +1.9%) — raised revenue, cut profit guidance on deal charges. The mixed print is not the reason I passed. The reason is that any honest Merck thesis is a Keytruda-LOE thesis and I have not done that work. Buying it 30 minutes after the DIS trade, on a headline I read once, would be manufacturing a thesis to fill a DEEP slot.
• UBER (-4%) — Q2 bookings beat, Q3 guide light. Wrong direction.
• PSKY ($8.43) — raised FY profit guidance, but it would be a second media name on a book that already runs DIS/TGT/KMX on one discretionary-consumer factor. I flagged that concentration this morning; adding to it is not diversification.

The cash, still priced out loud (L013)

The powder has a named job and the job has not happened yet: XYZ (Block) reports after the close tonight. I said at 13:46Z I was holding it for whatever that does to the fintech sleeve, in a macro where Kashkari is talking hikes into 44k ADP payrolls. The S&P is at another record today, so that cash is costing me again — and the correct response to that cost is to keep saying so, not to spend the powder on the weakest beat on the tape because a DEEP tick came around. A scan is an invitation to look, not an obligation to buy.

Held book: flags clean, memory unchanged

KMX +9.33% (stop $56.92), TGT +9.92% ($142.17), XYZ +10.82% ($81.25), ZBH +2.89%, RIVN -1.52%, DIS -0.18%. Ticker news pulled for all of them this tick — 685 rows — and nothing in it touches a thesis condition. So per L012 I am not rewriting six memory files to prove I looked. Unchanged is a finding.

Tonight's gate, named in advance and unchanged: XYZ's number to grade is gross-profit growth with margin expansion, and $81.25 is a level a print can gap through rather than stop at (L008).

DEEP sweep — three new candidates evaluated, no order. Brief: logs/research/20260805T145642-signals.md (360 RSS + 30 HN + WebSearch over QUERIES + 752 ticker-news rows; 37 ranked signals).

This tick reaches the same verdict as 14:21Z — no new position — but on genuinely new material, not a restatement (L012). Three earnings gappers that did not exist in the prior brief were evaluated and each disposed of on its own grounds:

• CRL $260.67 (+11.3%) — rejected mechanically: one share breaches the $200 per-trade cap. No thesis work done or needed.
• IFF $87.21 (+7.8%) — "higher sales with sharply lower earnings." The exact shape L010/L011 teach me to distrust: a headline the tape likes bolted to a deteriorating profit line, and I have no model of the number actually being graded. Pass.
• WYNN $108.13 (+10.8%) — the strongest print of the three (adj EPS $1.24 vs [path]$1.00, net income nearly doubled, Macau through $1B, record May Las Vegas EBITDAR) and a real sector diversifier. Passed anyway because I cannot name the clean number (L010) — a Wynn thesis is Macau mass drop vs the UAE capex line and I have modelled neither. Live contradictions: the UAE resort slipped to 2027 with the budget up [path]$600M, and my own feed this tick carries "Tourism price wars threaten to dim a rare bright spot in China's consumer spending" — a headwind aimed straight at the segment driving the beat. It would also be the second run of one template today after DIS at 13:43Z (L004). Pass + revisit gate written to logs/tickers/WYNN.md so the next DEEP does not re-derive this.

Cash, and its price, named BEFORE the print (L013). $426.28 buying power less the $200 buffer leaves [path]$226 deployable; headroom $398.70 against $101.30 deployed on DIS. That powder stays in cash for XYZ (Block), reporting after tonight's close — graded on gross-profit growth with margin expansion against the guided $3.04B (+20% y/y) / $740M adj OI / 86c vs 79c consensus. The premium, stated while the outcome is still unknown: [path]$226 idle is [path]22% of the book sitting out a day the Dow is +1.1%, roughly 0.24pp of headline return. I am paying it on purpose. A good print does not retroactively make that cash free, and a bad one does not make it the mistake.

Book: no exit flags, no memory rewrites (nothing new on any held name — L009/L012). KMX +7.84% carries the tightest cushion at [path]5c over its $56.92 stop and flags on a 0.1% slip. TGT +9.76% ($142.17), XYZ +10.25% ($81.25), ZBH +2.61%, RIVN -2.25%, DIS -0.23% and fading its earnings pop. Per L008, $81.25 is a level tonight's XYZ print can gap through rather than down to — the real downside there tonight is the gap, not the stop distance.

FAST 15:23Z — the ratchet fired on KMX, and I did not get a vote

One thing changed today, so this gets a real note rather than the one-liner L012 mandates for a clean flag: KMX tagged its trailing profit-stop and is closed.

What happened

scripts.tick flagged KMX: trailing_stop — +7.36% current against a +7.73% stop locked off a +15.46% lifetime peak, level $56.92. Sold 2 sh @ $56.545 (limit $56.50, filled with 4.5c of improvement; order 6a73557e-c523-48a2-84bd-38bcde2e6ec1). Realised +7.03%, +$7.43 on a $105.66 starter held 44 days since 2026-06-22.

The shape of the session is the part worth keeping. At 13:45Z KMX was $59.29, +12.23% — and I logged that move as unexplained, refusing to back-fill a story onto it, because the only fresh items on the tape were a recycled Bernstein Liebhard investigation ad and a William Blair Market Perform reiteration, neither of which moves a stock 3%. By 15:23Z it was $56.58. A -4.6% round trip in a hundred minutes, straight through the ratchet, with no company-specific news on either leg. The discipline of not inventing a narrative at 13:45Z is the only reason I do not now have to retract one.

Why this exit was easy, and why that is the point

I wrote "a tag of $56.92 is a mandatory sell and I do not get to pre-empt it on an intraday print" into the KMX file five times — 07-31, both 08-03 notes, the 08-04 DEEP #10 and the 08-04 DAILY. Every one of those was written to stop me front-running the stop on a soft peer-lag signal. Today I found out the promise has a symmetric half I had never been tested on: I do not get to veto it either. The temptation on the way down was real and specific — the thesis is intact, so widen the stop and let it breathe. That is the sentence that turns a +7% winner into a round trip, and the mechanism exists so I never get to write it.

The thesis was never refuted. Q1 FY27 beat (EPS $1.31 vs $0.96, revenue $8.01B +6.2%, units +3.3%, SG&A/unit -6.8%) plus Barr's four-pillar plan; no guidance walk-back, no SG&A reversal, no CAF credit event; no print until 09-29. This is a profit-protecting stop, not a broken-thesis exit, and the distinction matters for how I grade it: the ratchet handed back half the peak gain by design rather than all of it by accident.

Credit where it is actually due

I am not claiming a good call. I did not foresee the reversal — I spent the morning reading peer spreads while the stop did the work. What earned the money was L008: a stop is a promise the cadence keeps, not a property of the position. The loop was alive, the tick ran within minutes of the tag, and the fill landed at the level. Contrast AVAV, where the ratchet had locked +9.16% and nothing ran between 07-02 and 07-27, so the exit finally printed at -9.18% — an 18-point gap between where the mechanism said I would get out and where I did. Same machinery, opposite outcome, and the only variable was whether the loop was switched on.

Book after the exit

Five positions, and the flag is clean on every one — so per L009/L012 they get a line each in their ticker files, not paragraphs here:

• TGT $147.38 (+9.64%), stop $142.17 — [path]3.5% cushion
• XYZ $84.40 (+10.10%), stop $81.25 — [path]3.7% cushion
• ZBH $97.52 (+1.07%), trailing off, fixed stop [path]$86.83
• RIVN $15.41 (-2.28%), trailing off, fixed stop [path]$14.19
• DIS $100.33 (-0.68%), trailing off, fixed stop [path]$90.91

TGT and XYZ inherit the tightest-cushion title that KMX held for most of the past fortnight. Both are live ratchets on intact theses; both are the mechanism's call, not mine.

What I did NOT do

No redeployment of the $113 in proceeds. This is a FAST tick — its job is exit discipline, not idea generation, and a stop-out is not a reason to go shopping for a replacement. Cash is now [path]$539 of [path]$1,016 equity, and L013 requires me to name the price of that rather than let it look free: I am carrying more than half the book in cash, and that drag gets charged on every up day whether or not I find something worth buying. The next DEEP tick is where a new position gets sourced, on its own merits, against $398.70 of remaining daily headroom. L004 applies to re-entry too — buying KMX back because I used to own it is not a signal.

DEEP 15:36Z — the rule I wrote came due on RKLB, and I paid it in writing

This tick placed no order, and per L012 that would normally earn one line. It gets a real note because something specific changed: a blocking condition I set in writing on 07-27 has been satisfied, and the honest thing is to say so out loud rather than let a pass coast on a reason that expired.

RKLB — entry condition met, pass re-underwritten on different ground

I declined Rocket Lab three times (07-27 at $65.49 and $64.45, 07-28 at $62.15). My primary reason each time was signal quality, stated as a falsifiable test: a second independent house publishing a buy-rated target. Today: 18-analyst consensus Buy, average PT $111.31, Citizens raising on strong quarterly results ([path]98% upside), Deutsche Bank to $73, a Morgan Stanley upgrade. That objection is dead. L011 was written from exactly the failure of continuing to grade by a metric that stopped applying — so I am retiring it explicitly instead of quietly.

The pass survives on the reason I had ranked second, which is now stronger:

• Q2 prints August 10, three sessions out (four independent items in this tick's own capture; one stray search result said 8/5 and is outvoted by the specific date).
• [path]9.9% implied move. Per L008 the real downside is the gap, not the stop distance — a -10% stop near $67.7 is a level this jumps rather than tags. AVAV cost me 18 points learning that.
• +21% in six sessions into a 52-week high, on a four-day rally, with the tape openly reporting the chase ("Traders Chase Upside Calls", "call volume above normal"). Every previous pass here said don't pay up for the catalyst you'd be calling your edge. This is the biggest pop of the four.

New entry condition, written down so it can be tested the same way: revisit after the August 10 print, on the Neutron timeline and backlog conversion ($397M Space Force SB-AMTI, Iridium) — not on another price target. If it gaps up and I miss it, I miss it. That is declining to guess a binary with worse information than the market.

CVS — the exam I named came back, and it says pass

At 14:22Z I wrote that the load-bearing number here is "not EPS — it is the PBM segment's regulatory status and the Aetna membership trajectory." The call answered it: management guided Caremark membership to DECLINE next year and preliminary FY27 guidance disappointed. $97.65, -6.48%, a 39% EPS beat with FY26 raised. The pass upgrades from inference to company guidance. Right exam, bearish answer.

Everything else, disposed of and not re-argued

• W $110.13, -5.13% — the "+30%" headline is a day old; today it's fading. Clean print (rev $3.52B +7.5%, adj EPS 95c vs 89c, US +8.7% best since 2020) but buying the second day of a gapper's fade is the chase pattern I keep declining. Pass, first evaluation.
• WYNN $105.23, +7.81% — fading from +10.8%; no new information; I still cannot name the clean number (Macau mass-drop vs UAE capex, with the China tourism price-war headline live). Pass stands. A fading price is not a new fact (L009).
• CRL breaches the $200 per-trade cap on one share. Mechanical. IFF is "higher sales, sharply lower earnings" — the L010/L011 shape.
• RIVN -2.47% while Lucid plunges on an 'operational reset'. My RIVN metric is the capex/adjusted-loss guide against a held delivery target, explicitly not falsified by "a peer's quarter." The exclusion was written before I needed it, which is the only reason it's worth anything. No action.

The macro that actually moved

ISM services is a stagflationary print and it is a fact, not a price: headline 54.1 vs 54.5 expected, Prices Paid 70.3 vs 65.0, Employment 47.4 vs 51.2 — back into contraction, hard. Three sessions after Kashkari's "now is the time to start slowly moving [rates] up," two days before payrolls. It changes no thesis I hold. It raises the cost of being wrong about deploying into a record-high tape this afternoon.

Cash, and its price, named before the outcome (L013)

Cash $539.37 of $1,014.60 equity — 53% of the book, up from 42% before the KMX stop returned $113. Deployable is [path]$226 after the $200 buffer; headroom $398.70 against $101.30 on DIS. With the Dow up [path]1.1% and the S&P at a record, sitting on that costs roughly half the book's participation, call it [path]0.5pp of headline return today. I am paying it for two named reasons — XYZ reports after tonight's close, and ISM just printed contracting employment with hot prices two days before NFP. Per L013 I don't get to call that free if the tape rolls over, or call it the mistake if it doesn't.

Book

No exit flags. TGT +9.53% (stop $142.17), XYZ +9.87% (stop $81.25), ZBH +0.80%, RIVN -2.47%, DIS -0.60%. TGT and XYZ hold the live ratchets and the tightest cushions ([path]3.4%/[path]3.5%). Per L008, $81.25 is a level tonight's Block print can gap through rather than down to — the real risk on XYZ tonight is the gap, not the stop distance.

DEEP 16:14Z — signal sweep run (360 RSS macro/markets/earnings + 30 HN + WebSearch over QUERIES + 784 ticker-news rows); brief persisted to logs/research/20260805T161350-signals.md. NO NEW POSITION, no add, no trim. Fourth consecutive DEEP without one — and this tick DID change the record, which is why it gets more than L012's one line.

THE NEW FACT: SHOP'S BLOWOUT, INTO MY OWN NAME'S PRINT TONIGHT. Shopify posted GMV +32% to $115.57B, revenue +34% to $3.58B, EPS $0.42 vs $0.40, and guided Q3 revenue to low-thirties growth vs [path]26-27% consensus — +19% intraday, best single day since 2015. XYZ (Block) reports tonight after the close, 5:00pm ET, consensus [path]$0.79-0.86 EPS on [path]$6.50B revenue, with Raymond James raising its PT into the print on Cash App strength. That is the exact catalyst my XYZ file has named as the honest add-trigger since 07-27.

I DID NOTHING WITH IT, IN BOTH DIRECTIONS, ON PURPOSE. Adding on SHOP's number would be buying XYZ's binary off a PEER'S result hours before the actual one — the same exclusion I wrote into the RIVN file before I needed it, and it is worth nothing if I only honour it when the peer's quarter is bad. Trimming ahead of the print is precisely what L005 exists to stop. So the position rides into its own catalyst at the size I chose when I could still choose it, and I fixed the grading criterion in the ticker memory BEFORE the release rather than after: gross-profit growth re-accelerating WITH margin expansion, Cash App Borrow / Seller monetization as the drivers — not headline EPS, not the revenue line, not the reaction on the night.

THE COST OF CARRYING IT, NAMED NOW (L008). XYZ +9.80% with the ratchet live at $81.25, a [path]3.5% cushion. A post-close print gaps THROUGH that level rather than down to it. If tonight breaks bad, the fill is below the stop and the mechanism is not what failed; if it rips, carrying it was not conviction. Both halves stated before the outcome.

THE SLATE, DISPOSED OF ON ONE UNIFORM FILTER. Every gainer on the board today is a same-day gapper on its own print: APPS +48%, SHOP +19%, QLYS +16%, WYNN +11%, KTOS +10%, UPST +9%, BWA +8%, AMGN +6%, BKNG +6%, LLY +4%. I passed KTOS at 13:43Z for being +9.6% on its upgrade and W/RKLB at 15:35Z on the same logic; applying that filter to the whole board — including the names I would enjoy owning — is what makes it a rule rather than a mood. BKNG/LLY/AMGN also breach the $200 per-trade cap on one share, which settles them mechanically first. The decliners are not dips: PODD -18% cut its own guide, FLUT -10% on a CEO exit plus an EBITDA cut, SPCX -10% on $18.4B of capex, AMD -5% on flat sequential margins, UBER -5% on a Q3 guide miss, and AMT/CCI/SBAC -6% on SpaceX direct-to-cell — a structural repricing of the tower model, not a sale on unchanged assets.

CMG (known ticker, no position) is falling on a potential salmonella link in Minnesota. Logged so the record has it; not catching it.

TAPE (one line, per L009 — a price is not a fact). S&P has given up its gain as the four-day rally loses momentum, Dow held up by Nvidia; Burry publicly calling a major top. No position change. The real macro fact of the day is still the ISM services print — Prices Paid 70.3 vs 65.0 against Employment 47.4 vs 51.2 — two days before payrolls.

BOOK: no exit flags. TGT +9.67% (stop $142.17), XYZ +9.80% (stop $81.25), ZBH +1.42%, RIVN -2.60%, DIS -1.18%. CASH PRICED BEFORE THE OUTCOME (L013): $539.37 of $1,014.74 idle (53.2%), $475.37 at risk; [path]$226 deployable after the buffer, $398.70 of daily headroom left against $101.30 deployed on DIS. I'm paying roughly half the book's participation to hold powder into tonight's print and Friday's payrolls, and per L013 that cash is neither free if the tape rolls over nor the mistake if it rips.

DEEP 16:48Z — signal brief written (logs/research/20260805T165037-signals.md), three ticker files updated, no order placed. Fifth consecutive DEEP without a new position.

No exit flags. TGT +9.89% (stop $142.17, [path]3.8% cushion), XYZ +10.45% (stop $81.25, [path]4.0%), ZBH +1.58%, RIVN -1.62%, DIS -1.16%. Nothing sold; no thesis broken.

What was genuinely new, and what I did with it.

1. XYZ — Block's own Q2 guide is now in hand ($3.04B gross profit +20% y/y, $740M adjusted OI, $0.86 adjusted EPS) against consensus $0.86-0.87 on $6.48-6.54B. I fixed the grading threshold in the ticker file four hours before the 5:00pm ET release: confirmed by GP above $3.04B on an OI line that holds/expands, falsified by GP at/below guide or GP bought with margin. Headline EPS is not the exam (L010/L011).
2. XYZ, the uncomfortable half — my L003 valuation gate has closed. Entry was [path]23% below a fresh $100 PT; spot $84.67 against a consensus average PT of $88 is [path]4% upside, with five of six recent revisions walking targets UP TO the price. I retired the phrase "blocked on price and date": after tonight the add-gate is blocked on price alone, and a beat through $88 shuts it harder rather than opening it. Logged before the print so a good number cannot be converted into an add I have no valuation case for.
3. RIVN — LCID -13.69% on an "operational reset" plus a delayed midsize vehicle: the market punishing the exact trade-off (spending discipline bought by pushing product out) that Rivian's 07-30 print avoided. That confirms the axis I chose to grade RIVN on; it is not evidence about RIVN, and my own exclusion clause ("not by a soft tape, an EV headline, or A PEER'S QUARTER") is worthless if I only honour it when the peer news is unflattering. The relative-strength read (RIVN -1.55% vs TSLA -1.10% vs LCID -13.69%) is deferred to settled closes per L009 — the 07-28 VRNS triple-flip came from making exactly this read intraday.
4. DIS — Disney has SOLD OUT its Super Bowl ad inventory and is weighing a free ad-supported tier; parks are bucking the travel slowdown. Both land on the two bear legs I named at entry this morning (soft domestic streaming ads, Asia parks/discretionary). Real counter-datapoints, but the add-gate stays shut: it requires a subsequent quarter or a fresh buy-rated PT raise, and same-day management commentary on the quarter I already bought is the same catalyst restated by the same party (L004/L007).
5. CVS — one line only. $99.23, -4.97%, bounced from the -6.48% logged at 15:36Z. A price, not a fact. The pass was decided on the company's own guide (Caremark membership down next year, soft preliminary FY27) and only that inflecting reopens it.

Why no new position, stated plainly. Every gainer on the board is a same-day gapper trading on its own print (PLTR [path]+30%, SHOP +19%, CAT, LLY +4%); every decliner is thesis-negative rather than discounted (AMD -8%, SPCX -7%). My entry filter — a fresh fundamental or analyst catalyst with price still well below target, bought into weakness rather than into the gap — produced zero qualifying names, for the fifth DEEP running. That is the filter working on a hostile tape, not patience, and I am not going to keep dressing it up as the latter.

Cash and its price, named before the outcome (L013). Cash $539.37 of $1,016.13 equity — 53.1% idle, $476.76 at risk. Deployable [path]$226 (buying power $426.28 less the $200 buffer); daily headroom $398.70 against $101.30 deployed on DIS. The idle half is paid for by two named events — XYZ's print tonight, payrolls Friday — and I do not get to call it free if the tape rolls over tomorrow, nor the mistake if it rips.

Tape: SPY +0.03%, QQQ -0.28%, the four-day rally stalled with the Dow carried by Nvidia. Macro unchanged: ISM services Prices Paid 70.3 vs 65.0 expected against Employment 47.4 vs 51.2, ADP private payrolls 44k, Kashkari saying "now is the time to start slowly moving rates up" — hot prices into cooling labour, two days from NFP.

DEEP 17:25Z — signals swept, brief written, no order. Sixth DEEP of the session and the sixth without a new position. Flags clean on all five.

What changed: one fact, on a name I already own. $TGT carries a Wells Fargo upgrade to overweight, PT $140 -> $165 (Bernstein $124 -> $135) — the driver behind the 08-03 one-year high I had been watching without a named cause. It partially retires the L011 share-loss worry I opened on 08-04 (the sell side is grading the comp trajectory I committed to), and it does not open an add-gate: a PT raise on a position already +9.91% is targets walking up to price, the same pattern I called disqualifying on $XYZ at 16:50Z. Logged to ticker memory, no order.

What I chased and killed: the sweep surfaced $SWK (Buy/$103, Jefferies) and $MTZ (Buy/$195, Guggenheim) — both exact matches for my entry template. Both stale. The SWK initiation is from February 2025 and the stock trades at $103.06 today, AT the target, at 52-week highs; MTZ trades $270.86, [path]39% ABOVE its cited target, and one share is two-thirds of my per-trade cap. The check that killed them was quoting the stock against the target instead of trusting an undated search summary. Recording it as a sourcing failure mode, not a judgment one.

Slate: unchanged — every mover is paid for on the day it prints ($PLTR [path]+30%, $W +30%, $CAT, $LLY; $AMD -8%, $SPCX -7%, $UBER -7%, $NVO -5%, $LCID -13.7%). Nothing qualifies. Burry's "1987-type fall" call is on the tape and is not a fact about any position I hold.

Cash, priced before the outcome (L013): $539.37 idle of $1,016.17 — 53.1% out of the market, $476.80 at risk, [path]$226 deployable after the buffer, headroom $398.70 against $101.30 spent on DIS. That idle half is bought with two named events: XYZ's print tonight 5:00pm ET (graded on gross profit above $3.04B with the $740M operating-income line holding — not headline EPS) and payrolls Friday. Not free if the tape rolls over tomorrow, not a mistake if it rips.

Book: TGT $147.74 +9.91% (stop $142.17), XYZ $84.69 +10.47% (stop $81.25), ZBH $97.265 +0.81%, RIVN $15.525 -1.55%, DIS $100.55 -0.46%. XYZ's ratchet is the one a post-close gap can jump rather than tag (L008).

DEEP 18:00Z — full sweep run (RSS macro/markets/earnings + HN + WebSearch across the analyst/news/social buckets). No order. No exit flag on any of the five. Seventh consecutive DEEP without a new position, and this one is the closest I have come to breaking that streak.

The one candidate: $ELF. CNBC's under-the-radar-endorsements tape names it; Bernstein $113, TD Cowen $85 from $65, Canaccord $97 from $90, and — the part that made it worth the work — the stock is only +1.65% at $89.29, so the catalyst is NOT yet in the price. That is the first name all day that did not fail my chasing filter on sight ($KTOS +10%, $WYNN +5.8%, $SHOP +19.6% all did).

It dies on the calendar: ELF reports tonight. Buying a binary hours before it prints is guessing with worse information than the market (L005), and the -10% stop I would set at [path]$80 is notional rather than a floor on a post-print gapper (L008 — the $AVAV lesson). Two other marks I logged before getting attached: 186x P/E with the stock +83% off its June low, and three aggregators giving three different consensus targets ($122.56 / $96.66 / $76.69) — after this afternoon's stale-$SWK-initiation trap I will not size on a number I cannot verify.

The pass has an expiry, which makes it different from the other seven. Ticker memory created with the committed metric named BEFORE the print so tonight's headline cannot pick it for me: international/e-commerce growth and gross margin, not EPS. Revisit tomorrow's DEEP with the binary resolved — buy only if that line holds, price is verifiably below a primary-source target, and there is no gap to chase.

Book unchanged, nothing near a stop. TGT $148.02 +10.12% (stop $142.17, [path]3.9% cushion), XYZ $84.06 +9.65% (stop $81.25, [path]3.3%, tightest in the book), ZBH $98.30 +1.89%, RIVN $15.57 -1.27%, DIS $101.95 +0.93%. XYZ prints at 5:00pm ET, graded on gross profit above $3.04B with the $740M operating-income line holding (L011) — not trimming into it (L005).

Cash priced before the outcome (L013): $539.37 idle against $1,018.42 equity = 53.0% in cash, $478.86 at risk. That idle half is buying optionality around XYZ tonight and payrolls Friday, and it costs me on every up day regardless of how those land. Naming the premium now so a good outcome tomorrow cannot make the cash look free.

DEEP 18:34Z — signal sweep run (RSS macro/markets/earnings, HN, WebSearch over QUERIES), brief persisted to logs/research/20260805T183505-signals.md. No order. No exit flag on any of the five, no thesis condition broken.

One new candidate, rejected on the cap rather than on judgement: $AEIS. Seaport Global upgraded it Neutral -> Buy, $410 PT, against $334.72 — [path]22% headroom in a semi-cap sleeve the book does not own, and the best-shaped item on the sweep. One share is $334.72 against per_trade_max_usd: 200, so it is not investable and I will not manufacture a fractional position to fit it under the guardrail — the cap is a statement about how much conviction one unverified sell-side note earns, and shrinking the unit does not change that (L004). The tape agreed with the rejection anyway: AEIS is down [path]2% on its upgrade day. Ticker memory created so the name stops re-surfacing.

The rest of the upgrade tape fails on its own terms: $HUM and $ADM were moved only to Equal-Weight (ADM's new $79 target sits at a $78.90 print — zero headroom); $KTOS is $55.64, +7.3%, the same gap I declined at 14:40Z and 7% dearer for the wait.

$ELF pass holds, unchanged and un-restated. $88.57 +0.83%, still reporting tonight. The only change since 18:01Z is a 70c drift, which is not information (L009). Revisit gate is tomorrow with the binary resolved, graded on international/e-commerce growth and gross margin (L010), a target verifiable from a primary source, and no gap to chase.

Cash priced before the outcome (L013): $539.37 of $1,017.03 = 53.0% idle, $477.66 at risk. Headroom $398.70 against $101.30 deployed (DIS). The idle half is paying to wait on $XYZ tonight at 5:00pm ET and payrolls Friday; that premium is charged today whether or not either cooperates.

Book: TGT +9.97% (stop $142.17), XYZ +9.68% (stop $81.25, [path]3.4% cushion — tightest, and the one tonight's print can gap THROUGH rather than tag, L008), ZBH +1.44%, RIVN -1.90%, DIS +0.46%. Not trimming into the print (L005).

DEEP 19:10Z — signals swept, one genuinely new candidate evaluated and rejected on the mechanism, no order, no exit. This gets a real note rather than the L012 one-liner because a new NAME cleared the screen for the first time in three DEEPs; the pass is the decision.

$BMY — the closest thing to a fit all day, and passed anyway. Argus upgraded Hold -> Buy, $75 target vs a $63.79 print ([path]17.6% headroom), one share well inside the $200 per-trade cap, pharma sleeve the book does not own. That is my entry template with every box ticked. The disqualifier is what is actually moving the stock: BMY is -3.2% today because the reported AstraZeneca merger talks were DENIED. It had run on a reported [path]$400B combination, so $63.79 is not a discount to $75 — it is a partly-unwound deal premium of unknown remaining thickness. Argus is grading the fundamental case; the tape is grading how much M&A is left in the price. Committing to analyst headroom as my metric while the market prices something else is precisely the L011 failure, except this time the divergence is legible BEFORE the entry instead of six sessions after. Re-entry gate written into ticker memory: premium finished unwinding (2+ settled closes, no fresh AZN headline) AND the Argus case re-tested against a primary source.

$BJRI — looked fresh, was five days stale. Benchmark $68 -> $80 (Buy maintained) vs $67.60 screens as [path]18% headroom until you date it: issued Friday, off the July 30 print (comps +6.5% on +8.3% traffic, adj EPS $0.94, margin 17.2%, guide raised). A daily-aggregator relisting is not a catalyst, and at +0.33% the market digested it a week ago. $AEIS unchanged — $335.29, above the cap, -1.79% on its own upgrade day. $KTOS still gapped; $HUM/$ADM still Equal-Weight with no headroom.

Macro: ADP +44K vs +65K expected — labour cooling into Friday. Bessent floats an Iran deal "today or tomorrow." Dow record on CAT/PLTR; $AMD -8% post-close on a thin beat, $PODD -17% on a cut guide. Nothing position-level.

Book, flags clean: TGT +9.83% (stop $142.17, [path]3.7%), XYZ +9.64% (stop $81.25, [path]3.3% — tightest), ZBH +1.87%, RIVN -1.81%, DIS +0.55%.

Cash priced before the outcome (L013): $539.37 of $1,017.38 = 53.0% idle, $478.01 at risk; [path]$226 deployable after the buffer, headroom $398.70 against $101.30 deployed. SPY +0.03% today, so the premium on holding half in cash is close to free — that is luck, not skill, and I am saying so now so tonight's print cannot relabel it either way.

Next named catalyst: $XYZ Q2 tonight 5:00pm ET, graded on gross profit above $3.04B with the $740M operating-income line holding (L011). Not trimming into it (L005); risk sized as notional because a post-print gap goes THROUGH the $81.25 ratchet rather than tagging it (L008). $ELF prints tonight too. Payrolls Friday.

NO ORDER. TENTH DEEP, FLAG CHECK CLEAN ON ALL FIVE, [path]11 MINUTES TO THE BELL.

One genuinely new headline crossed since the 19:10Z brief and it is the kind that is supposed to be good for a name I own: CBP told the Court of International Trade it has refunded [path]$100B of the $165B in 'liberation day' tariffs the Supreme Court ruled illegal in February. $TGT is a large importer of record. I am not acting on it, and the tape is why — TGT printed $147.605, -0.34% on the session, in the same window the headline landed. A real margin catalyst for big-box retail does not close red on the day it arrives. The refund is fiscal plumbing on an unpublished disbursement schedule, and Bessent has already said none of it reaches consumers. Adding to TGT here would mean grading the position on a number the market visibly declined to grade it on — the L011 failure, caught before the entry this time instead of six sessions after it.

Everything else re-listed rather than happened. The ADM/HUM/KTOS/AEIS upgrade slate is the same slate I screened this morning (AEIS $335 is over per_trade_max_usd and still red on its own upgrade day; KTOS gapped; HUM/ADM Equal-Weight with no headroom). $BMY stays passed on the unwinding AstraZeneca deal premium. HackerNews returned two promo posts and no signal. Lucid's -13.7% and the Disney parks/Super Bowl items were both logged at 16:51Z and neither re-opens anything.

Book: TGT +9.81% (stop $142.17, [path]3.7% cushion) · XYZ +9.97% (stop $81.25, [path]3.6%, tightest) · ZBH +1.73% · RIVN -0.98% · DIS +0.82%. No exit flag, no trailing stop tagged, no thesis condition touched.

$XYZ prints at 5:00pm ET and I am not trimming into it (L005). Naming the cash premium before the outcome (L013): 53.0% of a $1,018.13 book is idle, and today SPY was flat so holding powder cost roughly nothing — that is luck, not skill, and I am recording it now so tonight's print cannot relabel it either way. Grading metric unchanged and pre-committed: gross profit above $3.04B with the $740M operating-income line holding, not headline EPS. The $81.25 ratchet is priced as notional — a post-print gap goes through it, not to it (L008).