Daily Report — 08/05/26 · “Monetized and Unhedged”

The market took profits all day and hedged almost none of it

Strip the day to its one load-bearing fact. Of every dollar of call premium sold across the whole 08/05 tape — $3.45B of it after removing duplicates and matched structures — 87% was struck in the money or within fifteen percent of spot, written against names that are up 9.4% on a premium-weighted basis over the trailing week. That is not bears opening shorts. That is longs ringing the register. The genuinely bearish share — far out-of-the-money calls, freshly opened, in size above existing open interest — was 3%.

The darkpool channel says the same thing by omission. Across more than three thousand names there was no campaign-grade block footprint on either side: 2,641 names printed zero darkpool prints, 651 printed thin single-cross transfers, and only 16 carried thick tape. Institutions were not moving stock on Wednesday. They were writing calls against it — the one form of profit-taking no darkpool screen can see, which is why both checks have to be quoted together: the block screen answers whether stock is being distributed, the moneyness screen answers whether winners are being cashed. Stock: no. Winners: emphatically.

And here is the part that matters for Thursday and Friday: they did not spend the proceeds on protection. Puts were net sold across the file, once matched legs and deep-in-the-money stock substitutes are removed from the aggregate. The only genuine put demand on the whole structure-adjusted tape sat in three places: SpaceX ahead of today’s unlock, the gold miners, and a Friday-expiry semiconductor hedge. So the market enters the payroll print monetized and unhedged — upside capped by freshly written call supply that dealers will lean on through August 21, downside cushioned by nothing in particular.

TAPE Whole file $17.79B gross after dedupe, structure stripped before any side was read. Sold-call moneyness: in-the-money $1.54B (45%) · within 15% of spot $1.46B (42%) · beyond 15% out, freshly opened above open interest — the only genuinely bearish bucket — $107M, 3.0%. Premium-weighted trailing move on the monetized names +9.4% (07/29–08/05). Darkpool campaign screen: no campaign named, either side (zero-print 2,641 · thin 651 · mid 54 · thick 16).

Call it what it is: not a distribution top, a harvest inside an uptrend. The distinction is tradeable. A top shows you block supply and put accumulation; this shows you overwriting. Overwriting is vol supply — it is why implied volatility fell across the whole surface on a red day — and it is fuel that burns off on a schedule. The lid the market built over itself this week dies at the August 21 expiration. Until then, rallies are sold by the people who own the most stock; after then, they are not.

Grading Monday’s report against two sessions of tape

The 08/03 report made twenty-four gradeable statements. Through Wednesday’s close: 9 hits, 4 misses, 2 partials, 5 still pending, 4 ungradeable. Strict hit rate on the fifteen closed items: 60%. Counting partials at half: 67%. That is a material improvement on the 27% the 07/31 report scored, and the composition inverted: this time the single-name and rotation calls carried the report, and the band-geometry calls sank it.

The worst miss was the central argument. Monday’s report said the weekly band would cap Tuesday, sketched a narrow, jumpy session between 7565 and 7650, and told you not to chase. Tuesday then printed one of the strongest sessions of the window — SPY +1.80% to 771.33, QQQ +3.40%, the S&P roughly 87 points through the stated ceiling. The report had itself conceded the weekly bands were struck off the 07/31 close and never reissued — and then leaned its strongest forward constraint on the surface it had just called stale. The lesson is procedural and it is now standing: a band that has already been exceeded is evidence about the band, not about the market.

The best call was the squeeze-fuel screen. All ten names on Monday’s list — MU, SKHY, AVGO, AMD, CAT, MRVL, SNDK, BE, LITE, and the semiconductor complex proxy — rallied on Tuesday, and every single one beat SPY’s +1.80%: MRVL +12.8%, SNDK +10.8%, LITE +8.9%, SKHY +8.2%, MU +7.6%, AMD +7.0%. The named sharpest candidate, SKHY, delivered the fourth-largest move. The rotation instruction — own the single names, not the index — was decisively right in relative terms, though the avoid-the-index leg cost absolute return on a day the index also rallied. The one caveat the screen carried, that the squeeze “needed the Friday catalyst,” was wrong about the ignition: the where was right, the when fired three days early.

Other closed items: MSFT “digestion, not reversal” — hit, the stock went +1.06% then −1.09% and ended two sessions within four cents of flat. AMZN “supportive” — miss, down both sessions against a rallying tape. The AMD earnings read — “a single-name event, not an index one,” with an 8.6% priced move — hit on both halves: AMD fell 7.04% inside the priced band and the index closed within a fifth of a percent of flat. The QQQ spread-arithmetic correction against a commentary target — hit within one session. The energy warning built on the 25,000-lot October put spread in the oil-and-gas producers fund XOP — hit, with XOP down 4.9% in two sessions straight toward the long strike. Still pending: the CRWV ceiling (untested, high close 91.90 against a 110–115 cap), the August structural ceiling, the Friday payroll resolution, and the Japan tail, which is graded after August 21 by design.

Dispersion is still the regime, and gold vol is the only vol anyone is paying for

For the second straight session the index barely moved while single names moved violently. GDX rose 7.39% and GLD 4.14% while AMD fell 7.04%, SpaceX fell 13.61%, and SPY closed down 0.20%. The VIX sat near 16 all day. Mid-session, the same eleven SPY contracts re-quoted an hour apart showed implied volatility marked down on all eleven while the index fell — a full red session that paid an at-the-money put owner nothing. That is not calm; it is correlation collapsing, with single-name variance netting out at the index level while the overwriting complex supplies vol on a schedule.

The regime has one kill-switch and the whole commentary layer independently found it this week: a correlation spike. A yen rescue that works, a call-squeeze unwind into late-August expiration, a rates accident — different doors into the same room, where every name moves the same way at once and the index vol that has been free all month suddenly is not. Until one of those doors opens, the payment schedule is unchanged: outright long premium bleeds, short single-name premium collects, and spreads that neutralize the vol leg are the only honest way to own direction.

Wednesday added one exquisite detail to that picture. Scan the entire volatility complex on the forward zone map and every equity-vol instrument sits crushed near the floor of its range — the VIX with 27.7% of upside room to its zone top, the nine-day index vol with 45.8%, the Nasdaq-100 and Russell vol gauges pinned at their lows with their trend readings reversed. One instrument on the whole board is bid: gold volatility, sitting at roughly 88% of its zone with 1.5% of headroom left. The only vol anyone paid up for on Wednesday was insurance on the one trade that has actually gone vertical. That is exactly where the smart hedging flow went — more on the miners below — and it is the cleanest single-image summary of this market: nobody fears the index; everybody who owns the metals rally fears giving it back.

The pin is confirmed four ways, and Wednesday closed on top of it

Four independent desks marked the same level this week without seeing each other’s work: Mike Silva’s deck puts the S&P gamma-flip at 7,505 with the market in dealer buy-the-dip mode above it; MAV’s zero-day walk called the SPY put-over-call inflections at 775 and 770 and both printed; a Tradytics options desk framed 767–773 as the “field goal range” with the overnight gamma pivot at 771; and Cem Karsan described index-level vol compression as the thing holding the whole market in place. Our own panel work agrees: the zero-day dealer book finished Wednesday with its largest positive gamma shelf at the SPY 770 strike — roughly $2.6B of stabilizing exposure — and SPY closed at 769.79, effectively parked on it.

The subtlety is what sits just above and just below. The strikes at 772 and 768 flipped mildly negative into the close, and on the S&P cash board the deepest zero-day negative-gamma pocket of the session sat at 7725–7735 — which is where the index closed. So Thursday opens inside a small amplification pocket wedged between large stabilizing shelves: moves start jumpy, then get caught. Layer the dealer calendar on top and the week’s shape writes itself. The August 21 monthly expiration carries the dealers’ biggest long book — roughly +$2.8B of net delta that dampens everything into it — while the Monday August 10 weekly book is the most negative row on the whole diary at roughly −$3.45B. Friday’s July payroll print lands two sessions before the most negatively positioned expiry of the month. A surprise in either direction on Friday gets amplified through Monday, and then the big August 21 book goes back to work catching it. That is the corridor: jumpy through Monday, pinned after.

Where price sits against its bands, and what each timeframe does to the next three sessions

The daily band computed off Wednesday’s close gives the S&P one standard deviation of room between 7676 and 7771 for Thursday. The upper edge of that daily band, the monthly ceiling at 7761.58, and the quarterly collar’s short-call strike at 7890 stack into the same overhead corridor — the index closed 38 points under the monthly line, which is half a normal Thursday. Daily says Thursday has ordinary room; monthly says the month has almost none. That asymmetry is the trade-shaping fact: every rally from here runs into the level where the market’s own August arithmetic and the collar’s standing call supply agree the air gets thin.

The weekly band tells you how stretched this already is: the sheet struck off the prior Friday put the week’s one-sigma top at 7594, and the index has now closed above it three straight sessions — about 129 points through on Wednesday. NVDA makes the same point in miniature: it closed at 219.22, above its weekly one-sigma top at 210.08 and within a single tick of the weekly two-sigma at 219.41. The gold miners went further — GDX at 83.68 closed above its weekly two-sigma at 81.68, the only instrument on the clean rows of the sheet to do so. The quarterly frame is the one that still has room: the S&P sits mid-band against a July–September range that stretches to 8069 on the upside, so nothing about this stretch is quarter-scale yet — and TLT, the only index-complex instrument that had breached its quarterly floor on Monday, closed back inside it at 83.00, healing the one quarterly-band violation on the board.

Put the four clocks together: the week is spent, the month is nearly spent, the day is ordinary, and the quarter is comfortable. Markets in that configuration grind sideways-to-up into their monthly ceiling and reject the first touch — which is precisely what the harvest flow above is positioned for, and why the extension case needs the one thing Wednesday’s tape sold: fresh call fuel.

The zone map: indices out of room, metals over the top, energy on the floor

The forward ZONE DOCUMENT and its Zone Visual trend layer — both computed off Wednesday’s close, both read in full — sort the entire board into three clean stories.

The index complex is pinned at its zone tops under intact uptrends. The S&P shows 0.38% of upside to its zone high against 5.37% of downside room; SPY 0.40% up versus 5.34% down; DIA is the extreme at 0.07% — seven basis points of headroom — and IWM has 0.59%. Every one carries a green trend line with the price dot pinned dark-red at the ceiling. This is the same exhausted-upward geometry Monday closed with, now two sessions more mature. The range readings that measure trend dominance are at cycle extremes across the complex — the S&P at 106, DIA at 121 — which means the trends are valid and fully priced at once: nothing about the zone map is bearish, and nothing about it pays you to chase.

The metals complex closed above its zone tops entirely. GLD’s printed upside to its zone high is negative — minus 0.70% — because the close at 389.64 sits over the ceiling. Silver’s is exactly 0.00%, the close and the zone high both printing 56.07. GDX at minus 0.85%, the junior miners at minus 0.90%, the metals-and-mining fund XME at minus 0.64%, palladium at 0.00%, gold futures with 0.45% left. Every dot in the cohort is pinned dark-red over the top of its range — and every trend line under those dots still renders red from the long basing period the breakout just exited. Price has outrun both the zones and the trend layer. Pair that with gold vol being the only bid vol on the board and the message is coherent: the people who own this move are insuring it, not adding to it, and chasing the pinned candle is buying what they are hedging.

Energy is the mirror image. The energy sector fund XLE sits at roughly a tenth of its zone with 4.9% of upside room, CVX near its floor with 6.4%, crude itself at the bottom fifth of a zone whose top is 22% away — and the oil trend readings are the invalid ones now: USO’s range collapsed to 5.7, which is a dead trend whose printed value carries no forward information, and the leveraged oil vehicles print outright reversed readings. Deep zone floors plus dead trend readings is the setup for violent mean reversion in whichever direction the next real catalyst points — it is not yet evidence the floor holds. XLE is also the one instrument on the whole board carrying a genuine multi-day distribution ladder: a cumulative dark net at a new low with a falling three-day slope. Energy is cheap in zone terms and still being sold in flow terms. The floor is a level to watch, not yet a trade.

Alphabet was the only mega-cap actually sold, and even that was profit-taking

GOOGL fell 4.03% to 362.43 and GOOG 4.05% to 360.13 — the researcher-departure headline landing on top of last week’s capex guide — and they were the only members of the mega-cap complex with genuine, citable selling in the options channel. Both share classes carry structure-clean negative residues that survived every stage of the decomposition, and GOOG’s is proportionally the larger and the more authentic of the two, with a real (if small) slice of freshly opened bearish call writing that GOOGL lacks. Watch the class gap: when the institutionally heavier share class leads the selling, it is positioning rather than headline reflex.

But grade the selling before extrapolating it: 92% of the call premium sold on GOOGL was in-the-money or near-the-money writing against a position that is up 13.2% over the trailing week. That is the same harvest signature as the rest of the file, executed on the one name with a fresh reason to take the money. GOOGL closed three cents above its 362.40 dark-tape demand shelf; the gamma map below it is bare down to the mid-340s and dealers are short, so a Thursday break of that shelf travels fast. Above, both classes have 5.8–5.9% of zone room — the most of any mega-cap except AAPL and TSLA — so this is a name where the map genuinely allows both outcomes and the shelf decides.

The rest of the magnificent eight: digestion, not distribution

Run the other seven of Silva’s mega-cap eight through Wednesday’s ledger and the same shape repeats — softness on the surface, harvest underneath, no exit anywhere.

NVDA was the day’s outlier and its biggest genuine buy: up 3.43% to 219.22 with a structure-clean options residue of +$90.3M, the largest single-name directional print in the file, on a residue that kept 71% of its gross after the structure strip — as clean as that ledger gets. Against it: the fifteen-day dark ladder still reads distribution with a falling slope, the close sits 0.30% under its zone ceiling at 219.87, and it closed one tick under its weekly two-sigma band. Flow says higher; the map says the move is fully priced through Thursday. The 81% of its call selling that was profit-locking says holders agree. With earnings three weeks out, the resolution is patience, not premium: no add at the ceiling, no overwrite against the file’s strongest buy print.

MSFT closed 487.46, a dime under the 487.50 supply shelf where $2.77B of dark tape sits, after its first red dark-net day since the earnings gap — on volume a third below average, with 96% of its sold calls harvesting the +25% post-earnings run. Its two large intraday blocks went opposite ways ($836M at the bid late morning, $274M at the ask mid-afternoon), which is two desks disagreeing, not a campaign. Dealers are short into the 480 shelf and buy dips. Digestion. META is the demoted one: price-based accumulation on the surface, but its day-count ladder tag was suppressed by its own falling cumulative net, its raw options tape sign-flips negative once structures are stripped, and supply is parked one tick overhead at 588.80. The burden of proof there is on the bulls. AMZN at 272.65 is controlled digestion of a +20% earnings run — a modest, genuine negative residue visible only after the structure strip, zero fresh bearish opening, and all three of its gamma magnets overhead at 277.5, 280 and 295, which argues drift up, not break down. AAPL is the one with room: up 0.52% to 311.00, sitting exactly on its own $4.22B demand shelf, with 10.5% of zone upside — the most in the complex — but its intraday sequence read “recovery sold into,” so the room is real and the sponsorship is not yet. TSLA at 321.55 printed a defended dip inside a two-week recovery, with the largest single dark band of its map at 319.70 directly underfoot — that remains the line. AVGO at 418.28 carries the single most internally consistent accumulation ladder on the whole board — cumulative net at a fifteen-day high, slope rising, 67% demand-side book, on a slow tape whose labels can be trusted — and closed pinned 0.96% under its zone top. It is the cleanest hold-and-do-nothing in the market: too sponsored to fade, too pinned to chase.

Rounding out the fifteen-name board the monthly sheet obliges this report to carry: INTC at 101.06 is constructive consolidation — its selling was 92% harvest against a +17% two-week run, supply was absorbed at the 100.90 shelf on the lowest volume of its window, and Monday’s inverted short call there remains a warning against fighting it. MU at 893.19 held flat and bought-dominant on reliable labels — still the preferred memory expression into any sympathy weakness, with long-dated call flow net positive into 2027. NFLX at 74.20 has no citable edge: its entire positive dark net was the closing auction cross over a down intraday tape, and dealers pin it in the 73–75 corridor. SPCX gets its own section below. MSTR at 98.37 sits mid-zone in its 90–100 coil with the gamma pin at 100 — the crypto section takes it up.

Semis and memory: the fight the whole commentary layer picked, and where the tape stands

The night’s twenty-one commentaries split cleanly down the middle on one question and agreed on almost everything else. The rollover camp — Karsan with a two-to-four-week momentum overhang from the positioning unwind, MAV short the semiconductor complex on vol-rank reversion, Mike Jones reading great-earnings-imploded as a completed pattern, FX Evolution watching the Korean market and the semiconductor fund top — says the complex rolls into late-August expiration. The defended-bid camp — Arete buying MU and the leveraged semis fund into the after-hours flush in real time, Silva keeping ARM on his list and treating the memory wreckage as a re-alert rather than an exit, and the standing $33M SK Hynix 2027 call position — says every dip keeps getting bought inside minutes.

Wednesday’s regular-session tape voted for the second camp before the after-hours prints hit. The options channel bought the memory complex against falling prices across four separate instruments at once: the semiconductor fund SMH +$21.3M structure-clean, TSM +$21.2M, the memory ETF DRAM +$20.7M, and MU +$19.6M — each one a readable residue, each one divergent against its own red tape, which is what accumulation into weakness looks like when it is real. SMH’s intraday fade was simultaneously bought in the dark — a $120M defense of the morning dip with both nearest shelves demand-tagged at 570.20 and 567.70. Against that stands AMD, the one genuine sell in the complex: −7.04% to 482.05 with a structure-clean −$58.9M residue, heavy overhead supply at 484.60, and — alone among the mega-caps — a monetization verdict that did not come back as harvest, meaning its call selling was not merely profit-taking. The complex is not rotating out; it is dispersing internally: compute bid, AMD sold, memory contested. Then SNDK and WDC reported after the close, beat, and fell 5.6% and 11.2% anyway. The whole question settles at this morning’s cash open: if the memory names get bought back inside the first hour the way Tuesday’s flush was, the defended-bid camp wins the week; if the semiconductor fund loses its 567.70 demand shelf, the rollover camp was early, not wrong.

Earnings: beats got sold again, and the dip map survived contact

The after-hours slate confirmed the absorption regime in its third consecutive reading: beats get sold unless they are clean beat-and-raise. WDC beat and guided fiscal-year earnings up triple digits and fell 11.2%; SNDK beat by 18% and fell 5.6%; HUBS beat by 28% and fell 20.1%; AXON beat by 24% and fell 6.6%; MELI beat and fell 4.4%; APP missed on revenue and fell 17.3%. What paid: PAYC +10.4%, SITM +25.1%, MSI, MCK, SOUN, LEU, ALB, OXY — every one a raise attached to a beat. In a tape like this, pre-positioning long premium into prints is donation; the play remains post-print, at the sponsored shelf, after the event vol crushes.

The dip map written Tuesday night survived contact with the census. WDC 430–460 stands — the fifteen-day accumulation ladder is still rising at +$3.4B cumulative — but it picked up the one genuinely bearish overlay in the whole earnings cohort: 64% of Wednesday’s sold-call premium on WDC was far out-of-the-money, freshly opened, above open interest. Somebody was writing real upside caps into the print, not just harvesting. That keeps the zone tradable and not yet investable: first bounce yes, conviction size only after the slope re-accelerates and that overhead writing stops. AXON upgraded on its own numbers: five consecutive buy days into the print, cumulative dark net at a fifteen-day high, slope rising +$273M, 78% demand-side book — the after-hours give-back to the low 570s is exactly the sponsored dip the map wanted, with the 600 strike carrying the gamma pocket that must hold. SNDK remains scalp-only and gated on MU: its options file was 95% structure — unreadable as direction — and the census shows no ladder and a supply shelf exactly at the 1350.50 close, so the only anchors are the event straddle’s implied exhaustion band in the low 1200s and the deep institutional floor near 1050 where January-2027 put sellers committed. The avoids all confirmed: HUBS carried the only clean bearish directional residue on the dip-map roster — $6.1M of fresh August and December put buying against a decaying ladder — before it fell 20%; MELI rallied 1.8% straight into a $228M supply shelf at 1922.60 with a falling cumulative net, the clearest distribution-into-strength profile on the board; APP’s February-2027 risk-reversal floor at 300 is where that conversation restarts, not before.

Thursday morning’s slate, positioned before the prints

The pre-market brief written overnight stands; the two that matter to the book: CEG — the nuclear-sleeve holding — goes in with patient money long (a June-2027 bullish risk-reversal plus a near-delta-one September put sale) and negative gamma that overshoots; a guidance-noise dip into 248–255 is the sponsored add zone, with the institutional deep floor at the sold June-2027 220 put. KVUE is the quiet sponsored long of the morning: an emerging accumulation ladder rising +$142M into the print, positive gamma, a $139M demand shelf at 19.50 one percent below spot, and — notably — zero options prints on Wednesday, so the event is being carried in shares. In this absorption regime that profile is the one that gaps and holds; a sold beat that returns to 19.50 is the entry, not the accident. DDOG is the bearish setup of the morning — a −$1.2B distribution ladder with a falling slope into the print, near-dated downside bought at 119–221 vol while the long-dated floor was written at 230 — the anti-WDC: its dip is not buyable until the slope turns, whatever the quarter says. U carries the widest implied move (roughly ±14%) and this slate’s HUBS-pattern candidate: $2.1M of front-week 33-strike puts swept at the ask in the final forty minutes at 243–270 vol. Respect whichever side that late buyer was on. FISV, TRGP and CELH all go in with falling-slope distribution ladders — no dip-buying claims on any of them.

SpaceX: the unlock is today, and the crash came first

Wednesday’s −13.61% SPCX break to 108.27 on 80% expanded dark volume now has its calendar fact: the first lockup expiration is today, Thursday August 6 — roughly 911M shares, about 20% of restricted holdings, eligible two trading days after Monday’s first-ever earnings print, with further staggered tranches behind it and the full expiry in early December. The three conflicting dates the commentary layer carried are settled; the thesis ledger’s old August 11 date was wrong and is corrected on the record. The tape into today was unambiguous: SpaceX showed one of only three pockets of genuine put demand on the whole market, its selling was front-loaded to this week’s expiry, and after a 13.6% gap-down the dark tape shows no mapped demand shelf underneath — the nearest structure is overhead resistance at 108.30. Supply meets an unpadded book today. The operator’s short-side structures are positioned for exactly this; the squeeze branch stays covered through the August 14 hedge. One flow footnote for honesty’s sake: the at-the-ask labels that dominated Wednesday’s fast crash tape are unreliable by construction — the price-and-volume read governs, and it says distribution on expansion.

Mike Silva, multimodal: the deck confirms the regime, number for number

Silva’s Wednesday report — forty-three slides plus the transcript, read as one input because most of the deck’s numbers are never spoken — is the closest thing to an independent audit of this framework’s regime read that exists. His board: the S&P at 7,723.55 with monthly momentum deep in overbought; the gamma-flip at 7,505 with the market comfortably above it, which is his dealers-buy-dips regime; the SKEW index collapsed to 133 from roughly 150 — downside skew imploding while the index made highs; the three-month-to-one-month vol ratio at 1.199, sitting exactly on the 1.20 line where his historical work says pullbacks tend to start; bullish-percent readings on the S&P, Nasdaq and technology at 70–82 with their momentum gauges at 70; the dollar down hard from 101.8 to 99.69 and the yen intervention marked as the driver of the ramp. His spoken conclusion: the tape is bull-strong, record S&P call volume squeezed the market off a long pin at 7,500 into new highs, skew imploded — expect digestion, take some profits into strength.

Overlay that on our ledger and the agreement is close to line-for-line. His “spot up, vol up, then skew implosion” is the same event our surface work logged as implied vol marked down across all eleven re-quoted contracts; his gamma-flip regime is our dealers-long-gamma pin at SPY 770; his overbought-breadth-plus-1.20-contango digestion call is our harvest verdict wearing breadth clothing; his “take some profits” is what 87% of Wednesday’s call selling was already doing. Two places to discount him: his swing playbook expresses everything in long delta and bought breakouts, which leaves the dispersion edge — the short-vol side his own numbers argue for — on the table; and his regime call (bull strong, carried as the standing overlay since Wednesday) is a commentary-layer input that sharpens the framework’s read without outranking the flow hierarchy. Where he is genuinely additive: the 1.20 vol-ratio line is a dated, checkable digestion trigger, and his watch list — all seventeen names of it — is ranked against our ledger at the bottom of this report.

MAV: he traded our tape from the other side of the screen

MAV’s nightly recap describes Wednesday almost exactly as our panels drew it, from a different desk with different tools. His core read: the call-side gamma squeeze died in one session; the vol skew that had been paying calls is correcting; SPY’s calls-to-puts ratio flipped from two-to-one toward puts; and Friday’s weekly expiry carries much bigger put open interest than calls, which he reads as downside risk into the jobs number — a claim this report holds for the structure test, because a fat weekly put line is one undecomposed leg of something until the matched-leg and financing checks run. His zero-day walk called the QQQ put-over-call inflection at 725 targeting 720 and the SPY inflection at 775 targeting 770 — both printed, in his words “720 and then some, 770 and then some,” and he booked a claimed $50K as QQQ closed below the 720 leg of his call-credit spread. When a retail desk’s inflection map and an institutional decomposition ledger mark the same four strikes on the same day, the strikes are real.

His trades are the overwriting supply complex operating in miniature — and mostly the right side of our tape. Selling hyperscaler calls: an MSFT 500/545 call-credit spread plus a 480/470 debit spread into mid-August, GOOG toward its long-term average on the executive-departure story, META on its breaking support — that is him joining the same call supply our monetization screen measured, on the same names. His structures are disciplined: credit spreads and body-debit-wing-sale combinations rather than naked premium. Where we fade him: the broad semiconductor short — his vol-rank reversion logic is a magnitude argument, not a direction argument, and it runs into the four-instrument memory bid documented above; his read of Friday’s big put open interest as a directional wall skips the structure test (large put lines at weeklies are routinely one leg of financing or collar structures — the same panel illusion that manufactured a false fortress in July); and his GOOG “negative cash flow” line is a factual error — Alphabet’s free cash flow is strongly positive; the real number is the capex guide. His unusual-activity flags — the Korean fund put sweep above all — are cross-checked against our own prints in the unusual section below.

Arete: the only real-time tape in the batch, and his bid is the question of the morning

Arete ran the SNDK and WDC prints live, and his session notes are the defended-bid camp’s best evidence precisely because they are observational: single names dumping violently after hours while the Nasdaq futures held and then rallied; semis “fell out of bed for three minutes and now you already have a bid”; he bought MU on the four-o’clock dip and added SOXL, the leveraged semiconductor fund, into the flush. His WDC puzzlement — a clean beat, raised guide, better margins, sold 11% while Seagate ripped on its own print — resolves through the same lens as everything else this week: positions that ran hard into a print get monetized through it regardless of the quarter, which is precisely what he independently concluded about PLTR (“crushed the quarter, no follow-through — an over and a reversal”). His APP read — margins soft, guide low, “something people will be shorting tomorrow” — printed −17% within the hour.

Two of his calls we fade, both on discipline rather than disagreement. His tactical gold sell signal off the two-o’clock rollover fights a structural breakout that our flow work grades as hedged-not-distributed — fading the metals complex on an intraday wobble while it closes above its zone tops is a timing trade against a regime. And his SHOP puts — bought because they looked inexpensive into the print and cashed for “a buck” — worked as a lottery ticket on event vol that happened to pay; buying the wing into an earnings print is the expression this regime punishes on average even when a given night pays it. Grade his real thesis at the open with everyone else’s: he is the one who was long the memory dip in real time.

Geeks of Finance: valuation bands against the flow

The Geeks piece is a fundamentals valuation pass on three reporters — no flow, no positioning, no vol surface — and it is useful precisely as the layer our tape cannot see. Their AAPL number: fair value 247.93 against a 309.38 close at the time, roughly 25% overvalued on their bands, both revenue and earnings softer year-over-year, offset by a two-to-three-year 500–600 model if the AI-infrastructure thesis lands. Their MSFT number: fair value 515.18, which puts the stock slightly under fair value even after the earnings gap — the better-timed of the two in their framing, with insider and congressional buying in the file. Their ServiceNow number: fair value 147.39 on NOW, basing near 100 — no position in our tape either way.

Where their model and our ledger touch, they mostly agree and differ on expression. Their MSFT undervaluation call converges with our accumulated-not-distributed read — but the flow adds what the model omits: MSFT sits inside the overwriting complex, so the paid expression of that agreement is monetizing the base through sold calls rather than chasing the candle. Their AAPL overvaluation is a two-year statement colliding with a tape where AAPL just held its demand shelf with the most zone room of any mega-cap — a valuation band is not a short trigger in a pinned-gamma regime, and the flow’s message on AAPL (room without sponsorship) is compatible with their message (quality without a discount): both say wait. The general rule stands: model fair values are not levels the tape respects, and on any conflict the ledger wins — but as a filter on what to do at the framework’s own levels, the valuation layer earns its place in the stack.

What the commentary layer agrees on, and the one place it splits

Thirty-seven commentaries across two batches this cycle, and they compress to four points of near-unanimity. One: the regime is dispersion — five unrelated desks rebuilt it independently from options vol, dealer gamma, breadth, macro issuance and tape parables. Two: the pin is real — four independent derivations of the same SPY 770–771 / S&P 7,505-flip structure. Three: the seller is structural — Karsan named the overwriting complex, an income-fund survey catalogued it, MAV traded alongside it, Silva measured its footprint in the skew, and our monetization screen counted it at 87%. Four: the kill-switch is a correlation spike, reachable through a working yen rescue, a call-squeeze unwind into late-August expiration, or a rates accident — and its cheapest tell is the vol-surface ratio between the Nasdaq-100 and the S&P compressing from its stretched 1.89x toward normal. The one genuine split is semis-versus-memory, documented above, and it settles at this morning’s open. The unanimous trade — sell front vol through spreads and calendars, own the body rather than the wing — indicts the two outright wing purchases still on the book and the $33M SK Hynix outright-call bet alike: right views, expensively expressed, in a market that pays the other side of that expression.

Unusual structures

The mechanical scan surfaced eleven structures worth taking apart. The theme across nearly all of them is the same as the market’s: professionals fencing profits and restructuring exposure, almost never opening naked direction.

ADBE: the day’s biggest “call buy” was a conversion, and the real story is a LEAP unwind

The largest non-mega-cap print of the session — 10,000 October 200 calls, $63.6M, crossed at 15:36 — looks like a monster bullish bet and is nothing of the kind. The same second carried 10,000 October 200 puts, and the package priced within pennies of exact put-call parity against a matching one-million-share darkpool cross at 258.25. That is a conversion — financing plumbing with zero market-facing delta. What survives it is more interesting: roughly $11M of deep-in-the-money 2027–2028 LEAP calls sold at the bid through the day — a long peeling about $23M of synthetic exposure into ADBE’s +20% two-week rebound, at 97% harvest-grade moneyness, while put sellers underwrote the 190–230 floor at flat skew. Read: the rebound is being cashed, not chased, with a supply shelf at 262.20 overhead. Watch the October 200 line’s open interest today — if both sides print up 10,000, the conversion is resting structure; if not, it was a transfer.

WYNN: a closing roll that got bigger, and the quiet bull of the day

At 14:40, in one second: 16,500 September 90 calls sold at the bid — 99% of that line’s entire open interest, a position closing — against 18,000 December 90 calls bought at the ask for $31.6M, fresh open interest. A post-earnings winner (WYNN +3.64% on the print) rolled three months out and upsized, paying up in premium and vol (39.7 implied on the December leg against 35.6 on the September being closed) to stay long a deep-in-the-money position through year-end. Institutions do not add size on a roll they doubt. Among the day’s larger structures this is the most straightforwardly bullish one on the board.

GDX: $42M of put spreads bought into a 7.4% rip — the smart money is insuring the metals run

At 11:03, with the miners already up big, one institution bought roughly 45,000 December 75 puts and 40,600 November 75 puts at the ask and financed part of the package by selling 44,900 December 67 puts — about $34M of premium paid, insuring roughly $685M of notional at the 75 strike. Two details give it away as hedging rather than a top call. First, the skew: the 75-strike puts printed at 43.5–44.0 implied against a near-money surface at roughly the same level — a 1.0x ratio, meaning the buyer paid no fear premium at all for downside on an instrument that just went vertical. Insurance was on sale because the overwriting complex crushes everything, and somebody with a large book noticed. Second, the rest of GDX’s tape: 97% of its call selling graded as harvest, its darkpool printed real breadth (25 prints on volume up 181% — one of the few names with genuine print-grade tape), and the intraday dark flow was net bullish once the closing cross is stripped — a price-and-volume read; GDX’s own options residue is thin, divergent and not the basis of any claim here. Owners banking and insuring a +7.4% day that closed above its zone ceiling — with gold vol the only bid vol on the board — is a coherent single story told three ways.

KRE: the regional-banks roll up and out, in size

Same-second, same-size, 55,933 contracts each way at 12:56: the August 78 calls sold at the bid, the October 80 calls bought at the ask for $11.9M at 22.3 implied — a matched two-leg roll — the same-second size pairing is the structure — extending two months and raising the strike two dollars on a line whose October open interest was a fraction of the print. A winner in the regional banks — KRE closed 77.34, high in a zone that tops at 78.17 — choosing to stay long through October rather than take the money. With the rate complex easing (the ten-year at 4.62%, the thirty-year back under its 5.20% line) the trade reads as a duration-sensitive financial bet extending, and it is the second roll-up-and-out on the day’s tape after WYNN. When multiple large books independently pay premium to push winners out in time, the message about their forward view is not subtle.

DRAM: the memory roll that doubled — hours before the memory prints

At 14:52, ninety minutes before SNDK and WDC reported: 26,400 August 53 calls sold at the bid against 52,800 August 60 calls bought at the ask — the memory ETF’s biggest structure of the day, a roll up seven dollars at double size, at 94 implied. This is the single most aggressive expression of the defended-bid thesis anywhere on the tape: someone took profits on in-the-money memory exposure and immediately re-risked twice the contracts higher, into the prints, at event-grade vol. The after-hours wreckage puts that position immediately under water — which makes it this morning’s cleanest tell. DRAM’s own dark tape ran a second straight down day; if the 60-strike buyer defends the complex at the open, the memory bid is real institutions, not hope.

LCID: $16M of 2027 puts on a $6.54 stock

Thirty thousand December-2027 5-strike puts and sixty thousand 3-strike puts, bought at the ask, freshly opened, at 134–136 implied — someone paid $16M for seventeen-month insurance against Lucid trading toward zero. At that tenor and those strikes this is not a trade on next quarter; it is solvency positioning — either a stake-holder hedging a locked position or a structural bear financing a long wait. Either way it is one of the largest capital commitments to a single-name downside view on the whole file, on a name with no offsetting bull tape whatsoever.

The Korea complex: MAV’s put sweep is real, and it is the memory fight in macro form

MAV flagged a large Korean-fund put buy overnight and our tape confirms it precisely: 40,000 August 120 puts on EWY — 42% out of the money, 41 cents each — alongside a structure-clean net of −$11.2M on the fund, the cleanest bearish residue among the day’s ETFs at 36% of its gross. Against it: SK Hynix’s ADR carried a positive structure-clean residue on a red day, plus the standing $33M of 2027 calls from earlier in the week. So Korea itself is the semis disagreement in miniature: index-level protection being bought in size while the flagship memory name gets accumulated underneath. Note what the put buyer paid: 42%-out-of-the-money wings at 41 cents are lottery pricing, and the sweep — timed with the Korean market down 4% overnight — reads as tail insurance bought at lottery prices on the exact correlation-spike scenario the whole commentary layer named, not a conviction short.

TLT: an 18-month duration bet, and the strongest buy-skew on the board

The long-bond fund printed the day’s most lopsided buying: +$12.2M structure-clean, driven almost entirely by January-2028 calls — nearly 49,000 contracts of the 90 strike taken at the ask, an 8.4% out-of-the-money strike eighteen months out, with the calls dashboard showing 83% of all TLT call orders on the buy side, the strongest skew of any liquid name. This landed the same session the thirty-year closed at 5.17%, back under the 5.20% line that our long-end framework had flagged as the kill-level for the “long end has stopped leading” claim — that claim is now retired by its own trigger. Somebody large is positioning for duration to work into 2027–2028, which is also the fiscal-dominance path: the accommodation trade, expressed patiently. The 83.20 supply shelf is the near-term test; the January-2028 open interest is the storyline.

The sector-hedge map: industrials underwritten, staples insured, semis crash-protected for one weekend

Three ETF structures drew the market’s hedging perimeter. In XLI, 50,000 November 170 puts were sold at the bid — $14.3M of premium underwriting industrials 9.6% lower, a willing-buyer commitment, not fear. In XLP, an institution rolled its consumer-staples insurance forward — selling the expiring September 78 puts and buying December 78 puts for a $7M net debit, keeping roughly $725M of staples notional insured through year-end. And in SOXL, 46,770 Friday-expiry 110 puts went up at 167 implied — a crash hedge on the leveraged semiconductor vehicle covering exactly the payroll print and the weekend, one of the three genuine pockets of put demand on the day. The composite: nobody is hedging the broad market; they are underwriting cyclicals, insuring staples through the election window, and buying one weekend of semis tail cover into the jobs number.

Smaller prints worth one line each

CAT: two 1,000-lot September 800-strike call blocks, $22M, deep in the money — stock-substitute accumulation in a name 66% up its zone; the industrial bid keeps showing up in structure rather than shares. CVX: a $6.8M September 165 call block, 12% in the money — same synthetic-long shape at the bottom of its zone, consistent with energy-floor positioning rather than chase. CI: a 5,000-lot October 270 straddle, $14.2M at 28–29 vol — somebody paying for movement in a health insurer with a 230% darkpool volume expansion and a −$395M label net; direction agnostic, event aware. COF: 10,300 October 220 puts a week after its zone read went top-of-range — card-credit protection at the 97th percentile of its zone. BKNG: 4,795 January-2027 at-the-money calls for $10.1M on a fresh post-split line — a patient re-entry in size. IONQ: 10,000 October 45 puts, in the money, opened ahead of its print — graded right, as the stock closed +2.1% after hours but the buyer owns three months. And the WMT headline flow that screens flagged all day — an $8.6M “clean” residue — was one 20,948-lot August 119 put sale at 91% intrinsic: a stock substitute wearing an options costume, quietly bullish, not a signal at its printed size.

Crypto: the ladder holds, the coin sleeps

Nothing in Wednesday’s tape moves the standing read. Bitcoin remains dormant below its 70.5K trend line, the futures-basis stretch remains a regime marker rather than a countdown, and the whole complex’s citable options flow stayed under $10M — there is no institutional stampede in either direction. IBIT is still the cleanest tape in the complex: a strong, slope-confirmed accumulation ladder at a fifteen-day cumulative high, +$88M on the three-day slope, with zero bearish call-opening against it — and it closed 36.74, mid-zone, inside a 35.5–37.6 band. The plan on the book is unchanged and has a Friday deadline: the expiring August 37.5 covered calls get re-issued as late-August or September 38–40 strikes — overwriting shares only, never the ladder — harvesting the pin while the add-tranche triggers wait at 34.2–35.4 below and the get-aggressive trigger waits at a weekly close through 37.8–38.3 above. MSTR closed 98.37 inside its coil with the dealer pin at 100 directly overhead and its options file too structure-choked to read as direction; the constructive detail was a patient buyer assembling December-2028 100/110 call spreads all day — defined-risk, 2.4-year paper — while 400 September 130 calls cleared at the bid at 76 vol, the same overwrite the book already runs. Both facts point the same way: the professionals express MSTR upside in spreads and sell its wings. The standing rule needs no restating beyond that.

The seventeen names on the desk, ranked best to worst

Silva’s Wednesday watch list — seventeen tickers, no levels, no notes — went through the full machine: census file, five-stage options decomposition, darkpool screen, monetization verdict, and a vol-surface check on every trade proposed. Ten of the seventeen have no census file, which is itself a finding: the framework can only vouch for what it can measure, and it says so per name rather than guessing. Ranked by evidence quality and actionability:

1. ARM — the list’s best structure: harvest the wing, keep the duration

ARM closed 274.58, down 2.13% the day after a 17% spike, and its tape is a museum-quality example of how institutions handle a winner. The morning sold near-dated calls into the pop — 90% of the day’s call selling graded as harvest against the +15% trailing run. Midday, the same tape bought roughly $3.2M of 2027–2028 calls at the ask, opening — including the March-2027 300 strike at 84.5 implied, the single most reasonably priced vol on ARM’s surface. Sell the front, buy the duration. The fifteen-day dark ladder is still rising (+$224M on the three-day slope) with a $211M demand shelf at 272. The trade mirrors the tape: against the long-term holding, sell the August 21 350 call at roughly 3.75–4.15 — that wing printed at 97.9 implied against an 88.9 at-the-money, an inverted call skew where the lottery strike costs more vol than the body, which is precisely the side to be short. The overwrite retires if the ladder slope flips negative and 272 breaks, or on a close through 300 with the LEAP accumulation continuing — then roll up, do not fight. Caution flag: the front week prints 108–136 implied, event-grade backwardation, so nothing gets bought near-dated on this name at all.

2. UNP — a billion-dollar accumulation ladder nobody talks about

Union Pacific closed 295.54, four cents above a demand shelf, carrying the strongest multi-day institutional footprint of all seventeen: $1.17B of cumulative fifteen-day dark accumulation, at a fifteen-day high, with the three-day slope rising +$336M — and Wednesday’s single opening options print was an institution writing 150 June-2027 255 puts at 28.5 implied, committing to own the stock 13% lower for ten months. Zero bearish-opening premium anywhere in the file. Positive gamma with the 305 strike as the overhead magnet. The trade joins the put writer with defined risk: September 285/275 put credit spread, short strike below the 295.50 shelf and the 290 gamma line — the visible September at-the-money printed 23.2–23.6 implied, unremarkable vol, so the spread’s edge is the ladder, not the surface; verify the short-strike vol at entry and stand down if it prices materially over that anchor. Exit on a daily close below 290, no averaging. Silva’s own transcript carries UNP targeting 302 then 315; the ladder is the reason to take that seriously.

3. KO — someone swept in-the-money calls with urgency, then built a collar around the year

Coca-Cola closed 86.83, up 0.31% on an accumulation signal, and the morning tape carried the most urgent prints on the entire seventeen-name list: 2,269 August 82.5 calls at the ask followed by 2,377 August 85 calls swept above the ask — high-delta, near-dated, aggressive. An hour later the same book laddered 2,900-lot call sales at 90, 95 and 100 across November, February-2027 and June-2027 and bought August 87 and January-2028 80 puts — a collar assembling around a long stock position, not a bear. The surface is dead flat at roughly 20 implied everywhere, so there is no skew edge to buy or sell; the edge is structural: buy the August 21 85 call (~2.43, 20.5 implied), sell the November 90 call (~2.80, 20.6 implied) — a diagonal that collects a small net credit, rides the sweep-buyer’s strike, and harvests the same November decay the 2,900-lot overwriter is harvesting. Hard calendar rule: at August expiry the short November leg does not stay naked — close it or convert to a vertical. Invalidation: a daily close below 85. Zone room to 91.30 gives the thesis 5% of headroom.

4. KVUE — the sponsored dip that reports in two hours

Kenvue is ranked this high on condition, not on conviction: it reports pre-market today, and Wednesday’s file had literally zero KVUE options prints — the event is being carried in shares, and its vol surface is unverified, so no pre-print position of any kind. What earns the rank is the profile underneath: an emerging accumulation ladder rising +$142M into the print, a $139M demand shelf at 19.50 one percent below the 19.67 close, no overhead supply band, positive gamma. In a regime where beats get sold and sponsored dips get bought, this is the sponsored profile. The plan is entirely post-print: if the reaction sells the beat and price holds or reclaims 19.50, buy the dip — shares, or a small cash-secured 19.50 put sold into the post-event vol crush once the surface is finally observable. A close below 19.50 on volume, or the ladder slope flipping, kills it.

5. MSTR — maintenance, not new risk

Covered above in the crypto section: hold the existing covered call, roll to the September 130 strike at ~2.35 (76 implied, printed at the bid in size Wednesday) when the August decays, and add nothing while the directional layer stays structure-choked. Wake-up trigger unchanged: Bitcoin reclaiming 70.5K retires the overwrite the same day. It ranks fifth because the structure is right and already on.

6. GH — two big dark sessions, one hedged insider-shaped put, verdict: watch

Guardant Health closed 163.18, up 2.12%, its second straight ~$150M at-ask dark session against a $20–85M base — a two-to-three-times volume regime shift with price up both days, which is how campaigns start. The options file is tiny ($400K) and leans the other way: a fresh deep-in-the-money September 180 put ($218K at 52 implied) is somebody paying real vol to protect a +15% two-week run, though the day’s last print was an October 160 call taken at the full ask. Two bullish inputs, one bearish, no ladder yet — below the conviction bar. No trade until either a third $100M+ dark session confirms the campaign or the earnings date is verified clear; then the expression is an October 160/185 call spread anchored to the printed 53.15 implied on the long leg — defined-risk, because a 52-vol biotech surface in a marked-down-vol regime is nothing to own naked. A close below 158 — under the block level — ends the story.

7. CSX — pinned, supported, and waiting for one print to resolve

CSX closed 51.12 on an accumulation signal, pinned between a $118M demand shelf at 51.10 and resistance at 51.60, inside heavy positive gamma centered on the 50 strike — a market going nowhere on purpose. The file’s one interesting print cannot be sided: 900 January-2027 50 puts, $252K, at dead mid, opening. Sold, it is a rail bull committing seventeen months out; bought, it is long-term protection. No trade until that line resolves — repeat prints at the bid make it a put-writing campaign worth joining via short puts; repeat at the ask is a de-risking signal. For an existing holder the September 55 call at 23.5 implied (~7.6% out) is a fair overwrite in a name whose call wing actually trades under its at-the-money vol — there is no upside premium being paid here to sell against, which is itself information: nobody is chasing CSX.

8. BMNR — a coordinated vol harvest on an 80-vol crypto proxy

No census file — every level here is print-tape context, not an anchor. What the ledger shows is coherent: $1.4M of call selling across every tenor scored 97% monetization-shaped, plus roughly 2,700 September 15 puts written at the bid, opening — premium sellers on both wings of an ether-treasury vehicle whose surface runs 77–95 implied while spot ground up 3%. Institutions expect range, not moonshot. The aligned expression, small and defined-risk only given the census gap and round-the-clock gap risk: September 15/12.5 bull put spread, short leg at the strike the informed writers chose, at their 81 implied — roughly 0.35–0.45 credit, with the long leg quoted live at entry. Selling 80-handle vol beside the professionals is a fundamentally different trade than buying it.

9–17. The rest: where the data ends, so does the opinion

TEVA (no census file): a $140K options file whose only coherent read is premium-selling on both wings — an at-the-money December put written at 38 implied, a 2028 45-strike call sold at the bid — range-harvest behavior, dust-sized; no trade. FBIN (no census): one post-earnings LEAP call sale at the bid, indistinguishable from covered-call harvest at 314 lots against 1,925 open interest; the residual front-week vol at 84.7 against 45.1 on the 2028 line is a steeply inverted term structure with no verified market on the short line — no trade. QURE (no census): one sliced ~$57K retail-scale buy of August 45 calls at 93–94 implied — event-priced vol on a binary gene-therapy name; buying that premium is the definition of anti-edge and there is no basis to sell it either — no trade. NWL, TAL, CORT, ANRO, BNED (no census, no options tape beyond isolated blocks): each one’s entire day was one or two darkpool blocks whose side tags carry no information at that print count — the honest description is institutional absence, and absence is not a setup. IRD is last by construction: the ticker prints exactly IRD on the source image, and there is no census file, no options print, no darkpool print and no ledger entry anywhere in the framework’s data — nothing exists to analyze. If the intended instrument was something else, one corrected letter reopens the question.

The prints MAV flagged, checked against our own tape

Scorekeeping the overnight flags, because his hit rate is becoming a usable input. The Korean-fund put sweep: confirmed on our tape to the contract — 40,000 August 120 puts on EWY, graded above as correlation-tail insurance at lottery pricing rather than a conviction short (41 cents for a 42%-out-of-the-money strike is lottery pricing, and lottery pricing is what tail insurance should cost). His hyperscaler call-selling structures land on the same names our monetization screen scored as harvest leaders — MSFT at 96%, GOOGL at 92% — so he is selling alongside the owners, which is the right company to keep. His SPY 775-to-770 and QQQ 725-to-720 inflection map printed both legs, and the 770 shelf he targeted is the same strike our dealer book shows as the largest stabilizer — independent confirmation of the pin from a third methodology. His weekly-expiry put-wall warning into Friday we discount on structure grounds (a fat put line at a weekly is one leg of something until proven otherwise), and his broad semiconductor short runs into the four-instrument memory bid — that one gets graded at today’s open with everything else. His ASTS 75-strike lottery calls and the TSLA 325/330/340 butterfly are trades, not signals: noted, not adopted.

BOTTOM LINE

The market spent Wednesday cashing a two-week winning streak through the options market — 87% of all call selling was profit-taking, the darkpool showed no distribution campaign anywhere, and puts were net sold. Direction is intact and fully priced at once: every index closed pinned at its zone top with the week’s band spent, the month’s band nearly spent, and the quarter’s wide open. The regime is still dispersion — the only vol bid on the entire board is gold vol, bought by the same institutions insuring the one parabolic trade in the market. The next three sessions have a defined shape: a payroll print Friday landing two days before the most negatively positioned weekly expiry of the month, inside a market that is monetized and unhedged, followed by a dealer book at the August 21 expiration that wants everything pinned. Buy sponsored dips at mapped shelves — WDC into 430–460, AXON toward the 600 gamma line, KVUE at 19.50 if the print is sold, MU on memory sympathy weakness — sell elevated single-name wings against stock, own index direction only through spreads, and let the semis-versus-memory fight settle at the open before adding anything in that complex. The one picture to keep from Wednesday: everybody took money off the table, and almost nobody paid for insurance. Markets in that state do not usually fall first; they stall, and then they move fast in whichever direction finally forces the crowd to pay up.

Top Trades to Follow

Institutional structures from the 08/05 tape worth tracking, graded in the next report’s scorecard:

LONG · ROLL WYNN — closed 16,500 Sep 90C (99% of the line’s OI) and opened 18,000 Dec 90C for $31.6M at the ask. A post-earnings winner extending three months and upsizing. Grade: does WYNN hold above 100 into September.

LONG · ROLL KRE — 55,933 Aug 78C sold / Oct 80C bought, same second, $11.9M. Regional banks extended through October on an easing rate tape. Grade: KRE above 78.17 zone top within three weeks.

LONG · DURATION TLT — ~49K Jan-2028 90C taken at the ask, 83% buy-side skew on the day’s call orders, the session the 30-year closed back under 5.20%. Grade: the 83.20 supply shelf breaks within five sessions.

HEDGE · METALS GDX — ~$34M of Nov/Dec 75-strike puts bought at the ask against 67-strike sales — a matched two-leg spread, not naked length — at flat 1.0x skew, into a +7.4% session that closed above its zone top. Grade: insurance, so graded on GDX holding 75 — and on whether flat-skew hedges keep getting this cheap in vol terms.

LONG · FLOOR-WRITE UNP — 150 Jun-2027 255P written, opening, at 28.5 IV, atop a $1.17B rising accumulation ladder. Grade: 295.50 shelf holds; 302 prints before 285.

LONG · FLOOR-WRITE CEG — Jun-2027 360C bought / 220P sold risk-reversal plus a near-delta-one Sep put sale, staged into this morning’s print. Grade: a sold beat that holds 248–255 confirms the sponsorship.

LONG · AGGRESSIVE DRAM — Aug 53C sold / 60C bought at DOUBLE size (52,800), 94 IV, ninety minutes before the memory prints that promptly went against it. Grade at today’s open: defended or abandoned. The single cleanest tell on the semis fight.

BEAR · SOLVENCY LCID — $16M of Dec-2027 5P and 3P bought at the ask, opened, 134–136 IV. Seventeen-month downside conviction with no bull tape against it. Grade: quarterly, not daily.

BEAR · TAIL EWY — 40,000 Aug 120P at $0.42, the correlation-spike tail expressed for lottery pricing, against a clean −$11.2M structure-adjusted bearish residue on the fund. Grade: worthless expiry is the base case; its value is what it says about who fears the yen channel.

HEDGE · WEEKEND SOXL — 46,770 Friday 110P at 167 IV: one weekend of semis crash cover spanning the payroll print. Grade: expires Friday; the print itself is the verdict.

WATCH · UNWIND ADBE — ~$11M of deep-ITM 2027–2028 LEAP calls peeled at the bid (97% harvest-grade) under a parity conversion headline. Grade: acceptance over the 262.20 shelf refutes the cap; a third LEAP-supply day confirms it.

CARRIED · BOOK ARM / KO / KVUE / IBIT / MSTR — the five operator-side structures constructed in this report’s ranked section; each carries its own invalidation line there and gets graded against it.

SOURCES

Every file opened this cycle, not a highlights sample. Three external URL fetches were made this cycle and are itemized below; nothing fetched overrides a flow or price figure anywhere in this report.

GATED LEDGERS · the numbers behind every flow claim The five-stage options decomposition of the 08/05 live flow CSV (1,214 symbols) · the darkpool ledger of the 08/05 market summary (1,354 symbols) · the monetization ledger (sold-call moneyness buckets, file verdict, per-symbol verdicts) · the weekly expected-move transcription, machine-validated with 17–19 defective rows quarantined and excluded · the session venue call sheet, checked for internal consistency against the day’s own structure map. Full chain run twice this session; every phase passed on the end-of-day data.

RAW DATA Live Options Flow 08/05 CSV (deduped on the standard signature before any read) · Darkpool Market Summary 08/05 CSV · options dashboard 08/05 PDF — 23 panel tiles, every tile read as an image · darkpool dashboard 08/05 PDF — 11 tiles, all read. Rendered at 150dpi with PyMuPDF.

EXPECTED MOVES · ZONES · SENTIMENT Forward set stamped 08/06 carrying the 08/05 closes: daily board (index/futures complex only — the artifact carries no single-name rows this cycle, declared, not substituted), zones render, range & trend sheet, ZONE DOCUMENT (10 pages) and Zone Visual (21 tiles), all read in full · the five 08/05-stamped views consumed live by Wednesday’s intraday sessions, from the archive · weekly sheet 08/03–08/07 (quarantined rows excluded: the sector funds XLK/XLF/XLE/XLI/XLV/XBI/XRT, MAGS, the garbled MSFT and NDX rows, QQQ’s wrong close, GLD/SLV inverted bands, and six futures rows) · monthly August 2026 board (no GOOG row exists on the artifact — declared) · quarterly July–September board · JPM collar sheet (strikes cited only; the capture is a July 1 snapshot and its overlays are stale) · FOM sentiment gauge 08/05, three pages: 69.8, GREED, five-day change +34.5 — the largest five-day gain in the tracked series, velocity trigger fired; tracker rows for 07/28–08/04 were never logged and are declared as a gap with two back-solved anchors · Mike Silva’s 08/05 trade watch list image (the seventeen names, transcribed at 4x zoom).

PER-TICKER CENSUS · 2026-08-05, 76 files opened and quoted AAPL · ADBE · ADSK · AMD · AMZN · ARM · AVGO · AXON · BABA · BKNG · BMNR(gap) · CAT · CEG · CI · COF · COIN · CRWV · CSX · CVX · DELL · DIA · DRAM · DUOL · EQT · EWY · GDX · GH · GLD · GOOG · GOOGL · HOOD · HUBS · HUT · HYG · IBIT · INTC · IONQ · IWM · KO · KRE · KVUE · KWEB · LCID · MELI · META · MSFT · MSTR · MU (wl2) · NBIS · NET · NFLX · NVDA · NVO · PLTR · QQQ · SHOP · SITM · SKHY · SLV · SMH · SNDK · SPCX · SPY · TLT · TMO · TSLA · TSM · UNP · USO · WDC · WMT · WPM · WYNN · XLE · XLF · XLI · XLP. No census file exists for: TAL, TEVA, CORT, QURE, IRD, NWL, ANRO, FBIN, BNED, BMNR, SOXL, UUP, PAYC — each declared as a gap where the name appears; no level was substituted from any other source. The 08/04 census was consulted for the scorecard grading layer.

COMMENTARY · 37 files across two batches, all read in full Night batch (21): Mike Silva 08/05 transcript and the matching 43-slide deck read as one input · Cem Karsan (re-transcribed via captions after a decode-loop) · MAV Patreon nightly · Arete Trading SNDK/WDC earnings live · ET Tradytics · Mike Jones · FX Evolution · Geeks of Finance · CitizenOfTheYear · Strong Man · Dividend Data · Allen Reminick · Andrei Jikh · Anthony Pompliano · Road To $1M · five tastylive pieces including the SK Hynix $33M call story and Michael Howell on global liquidity. Day batch (16): VolSignals · Trading Apologist · Darius Dale / 42 Macro · Crypto Tips · Benjamin Cowen · Conquer Trading · Krown · Digital Asset News · Motley Fool on AMD · James InvestAnswers on PLTR · Lines on Maps · Two Minute Papers · three tastylive pieces · the withheld Karsan fragment. All flow-graded in the two commentary analysis files before this report was written.

EXTERNAL URLS FETCHED · 2026-08-06 ISM Services July 2026 report (prnewswire, ismworld): 54.1, 25th month of expansion, Business Activity 59.1, New Orders 57.2, Employment 47.4 back in contraction, Prices Paid above 70 · SpaceX lockup schedule (investing.com, fool.com, purepowerpicks.com): first unlock today, 08/06 — ~911.5M shares, ~20%, two trading days after the 08/04 print; staggered ~7% tranches follow; full expiry 12/08; founder shares locked ~366 days. The thesis ledger’s prior 08/11 unlock date was wrong and is corrected this cycle.

PRIOR WORK GRADED OR CARRIED The published 08/03 daily report (graded above) · the 08/05 comprehensive analysis built this cycle · the pre-market earnings brief for 08/06 · the post-market earnings review and dip map from Tuesday night · the Robinhood tactical book deep-dive and IRA portfolio review · the rolling tracker at version 62 · the regime snapshot (0803 baseline, refreshed this cycle) · the thesis ledger · the correction log · and session logs 03 through 10 of 08/05 plus the overnight commentary batch log.

DECLARED GAPS · not substituted, not inferred The forward daily band sheet carries no single-name rows · the monthly board carries no GOOG row · thirteen watch-list and reference names have no census file (listed above) · IRD has no data in any channel · the FOM tracker’s 07/28–08/04 rows were never logged · MU’s gamma-strike map is suppressed for scale contamination and AMD’s strike levels are suspect (signs used, levels not cited) · NVDA’s earnings date (08/26 vs 08/19) and the Jackson Hole window remain unconfirmed across sources and are flagged rather than asserted.