Daily Report — 08/06/26 · “The First Invoice”

The register kept ringing, on a much lighter tape

The profit-taking regime did not pause to watch the wreckage. Of all the call premium sold across Thursday’s tape, 81% was struck in the money or within fifteen percent of spot — longs cashing a run that measures +5.1% premium-weighted over the trailing week. The genuinely bearish share, far out-of-the-money calls freshly opened in size, was 7% — up from 3% on Wednesday, a first uptick worth watching but nowhere near a directional statement. The darkpool screen stayed silent the same way it has all week: twelve names with thick tape out of more than three thousand, no block campaign either side. Stock is not being distributed; winners are still being cashed through the one channel no block screen can see. Both checks quoted together, as always: blocks say no, moneyness says yes.

What changed is the size of the tape itself. Options gross ran a third lighter than Wednesday, and the dark tape contracted 24–45% day-over-day on nearly every large name. A market that has already monetized does not need volume — and thin tape into a jobs-report Friday cuts both ways: less supply to absorb a rally, less demand to catch a break.

TAPE File $13.70B gross after dedupe, matched structures and deep-in-the-money stock substitutes stripped before any side was read (81% of the file survived the strip as genuinely directional). Sold-call moneyness: monetizing share 81.3% · far-OTM freshly-opened bearish bucket 7.4% · premium-weighted trailing move on the monetized names +5.1% (07/30–08/06). Darkpool campaign screen: no campaign, either side (zero-print 2,591 · thin 714 · mid 51 · thick 12). Individual harvest verdicts: PLTR against a +27.5% trailing run, AMZN +15.6%, ORCL +12.5%, MSFT, GOOGL, NVDA, IGV.

Grading Wednesday’s report against Thursday’s tape

The absorption call was right, and understated. Wednesday’s report said the after-hours earnings reaction was the tell of a market that cannot absorb good news; the marks got worse from there. Axon closed −14.3% against an after-hours print of −6.6%; Western Digital −13.0% on a clean beat; AppLovin −19.7%; SanDisk −6.8%. The dip-map discipline held its shape: Western Digital traded into the 430–460 buy zone named before the open and closed inside it at 451.52 — the test is live, not yet passed. The Axon read graded half right: the “sponsored dip” thesis said dark-pool accumulation would meet the post-print weakness, and the sponsor did stay — the fifteen-day accumulation footprint made a new high even on the wreck day — but the price damage was far beyond what any sponsor absorbed on day one. One tell failed outright: the memory-fund roll flagged as Thursday’s opening tell (the DRAM 53-strike calls rolled up to 60 at double size) died on arrival — DRAM closed −4.3%, below the lower strike. One failure of three allowed before the tell is retired; the clock is running. And Wednesday’s biggest clean buy print faded: NVDA, the largest structure-clean bullish residue of that session, went out flat at −0.1% and stalled exactly at the weekly band edge it had tagged to the tick.

What a market with nothing under it does to good news

Four prints, one slate, one mechanism. AppLovin −19.7%. Axon −14.3%. Western Digital −13.0% on a beat with a raise. SanDisk −6.8%. This is not four idiosyncratic disappointments — two of the four were unambiguous beats, and three of the four closed materially below their after-hours lows, meaning the selling continued into the liquid regular session. This is what the harvest posture predicted: when holders have already written calls against their winners and bought no downside protection, an earnings print — any print — becomes the exit window, because it is the only moment of guaranteed liquidity. The market cannot absorb good news at these prices because the people who own the good news have already sold its upside and have nothing forcing them to defend its downside.

The index never felt it: SPY closed −0.2%, pinned at the top of its range for a second straight session. That is the two-speed market in one line — the cap sits on the index while the invoice gets paid name by name.

Thursday’s own earnings: the sellers picked their targets, and software answered after the bell

The beat-selling extended to a third straight session — but for the first time it showed its criteria. Datadog beat cleanly on both lines and was sold −19.0% — the same treatment AppLovin, Axon and Western Digital got the day before, now applied to a fresh name on a fresh slate. Oscar Health beat enormously and fell −11.9%. Celsius missed and lost −18.5%. Kenvue, Constellation Energy, Keurig Dr Pepper and Howmet all reported fine-to-good numbers and closed flat-to-red. And then the other half: every energy and quality-defensive beat got paid — ConocoPhillips +1.5%, Targa +3.1%, Becton Dickinson +3.8%, Zoetis +3.9%, Cheniere +4.3% — and Unity, a growth name whose crowd left long ago, was paid +15.1% for its beat. The pattern is no longer “the market cannot absorb good news”; it is “the market cannot absorb good news where positioning is crowded and the multiple is rich.” Where the crowd already left, beats get paid. That is distribution-out-of-crowding, not risk-off — and it is precisely the shape the rotation footprint in the dark pools drew the same day.

After the bell, the countersignal arrived in size. The Trade Desk missed, guided a year-over-year revenue decline, and collapsed roughly −24% after hours to about $13.40 — on top of an 87% drawdown, putting it at the “down-90-percent” line where the deep-value crowd historically steps in. But around that wreck, the after-hours software slate was bought aggressively: Atlassian +23% (and roughly +31% by early Friday morning), Doximity +23%, Cloudflare +16%, Twilio +16%, JFrog +13%, Roku +10%, Airbnb +7–9%. That is the largest beat-paid cohort of the week gapping software up into a weekly expiration morning. Friday grades it: if those gaps hold into the close, the absorption regime re-reads as selective — miss-punishing, beat-paying — and the three-day negative streak ends. If sellers absorb a +20% overnight gap the way they absorbed Datadog’s beat, that would be the hardest confirmation of the distribution read this framework could ask for. Either way, the gap-and-fade question in software is Friday’s cleanest single-session experiment.

SLATE Morning prints, day-one closes: DDOG −19.03% (beat 0.65 vs 0.49) · CELH −18.46% (miss) · OSCR −11.86% (beat 1.10 vs 0.32) · KVUE −2.49% · CEG −1.52% · KDP −1.24% · HWM −0.58% vs COP +1.50% (paid, and unlike the pinned growth cohort COP sits mid-zone — roughly a third up its range with 4% of headroom on the forward map) · CNQ +1.56% · TRGP +3.12% · BDX +3.76% · ZTS +3.87% · LNG +4.32% · U +15.05%. After hours (unofficial, graded Friday): TTD −24% to ~$13.40 · TEAM +23% · DOCS +23% · NET +16% · TWLO +16% · FROG +13% · ROKU +10% · ABNB +7–9% · SG −15% · RKT −10%.

SanDisk: someone wrote $299M of panic at the bottom of the hole

The single most interesting flow of the day sat inside the ugliest chart. While SNDK fell 6.8% on nearly $1B of dark-pool distribution, the options tape printed the file’s largest structure-clean directional residue — +$336.9M net bullish, 2.4x the next name in the file — and its composition is what makes it professional: only $60.3M of it was call buying. $299.0M of it was puts SOLD, concentrated in this-week and next-week strikes from 1195 to 1265, several prints larger than the existing open interest, written at implied vols of 124–154 against back-month contracts on the same name priced near 100–107. That is not a bottom-fisher buying lottery tickets. That is size selling the most expensively-priced panic on the surface, at strikes below spot, with the earnings event already revealed — a risk-priced bet that the wreck is done, collected in premium instead of expressed in stock.

Discipline on what this does and does not license: the price signal won the day — distribution is the verdict, and a flow residue that disagrees with its own tape does not get to veto that. What the put-write does is mark where the floor-writers live: the dark tape’s own support shelf at 1214.80, $718M deep, held the close. If that shelf holds Friday, the put-writers are right and the wreck was the opportunity; if it breaks, they are wrong and the invoice has another page.

SNDK Close 1258.58 (−6.81%) on $972.67M dark volume. Options: $877.7M gross → structure and duplicates stripped → +$336.9M clean directional (38% of gross survived the strip — a genuinely high-conviction residue, flagged divergent-vs-price so it supports but cannot veto). Composition: calls bought +$60.3M · puts sold $299.0M (Fri/weekly 1195–1265 strikes, IV 124–154 vs ~100–107 back months — the sellers collected the event premium, they did not pay it). Dark support shelf 1214.80 ($718M).

Western Digital: the test arrived at 451, one buyer showed up at 458

WDC traded a 16% intraday range — a violent open fade to an 11:22 low, then a stabilization into the close at 451.52, inside the 430–460 zone the dip map named before any of it printed. The intraday dark tape was nearly empty — two prints all session — but the one that mattered was a +$48.6M buy at 457.98 at 3:09pm, a size print placed after the shape had resolved, exactly where a patient buyer executing the map would put it. (The −$298M closing-auction cross is excluded from that read — auction labels carry no information.) Two prints is not a verdict; it is a fingerprint. The rule from Wednesday stands: tradable, not investable — and Friday’s open grades it. Stability above 440 executes the thesis; a gap through 430 kills it, with no averaging into a broken shelf.

Axon: the wreck was retail-priced, and the sponsor didn’t blink

AXON’s −14.3% was the worst single day in its recent tape, and the strangest thing about it is what the institutional channel did: nothing. The fifteen-day dark-pool accumulation footprint — $562M cumulative — made a new high on the wreck day, with the three-day pace still rising at +$271M. Whoever has been building that position for three weeks did not sell the print. That doesn’t make the day bullish — a position that size underwater on a 14% gap is also how a forced seller gets made — but it narrows the question to one binary: either the sponsor defends the level Friday and the wreck was a transfer from weak hands, or the sponsor was simply late and the footprint starts shrinking. Watch the dark tape, not the price, for the answer.

Microsoft closed above its August ceiling — alone

MSFT +2.5% to 499.86 was the only mega-cap trend day on the board, and it did something no single name has done this month: it closed above its August monthly band ceiling — by five cents. The composition earns the skepticism the level demands. The dark-pool tape confirmed accumulation but on volume a quarter lighter than Wednesday; the options channel was nearly silent (a thin net-positive residue, half of it unreadable on side); and the profit-taking screen graded MSFT a harvest name again — the call-writing continued into the breakout. Price led, flow followed at a distance, and the sellers of upside never stopped. A close above that ceiling on expanding volume makes it a breakout; Friday closing back under it stamps this as the exhaustion print of a five-day run. The five cents decide nothing — the volume does.

Where price sits against its bands, and why Friday’s box is so tight

The four-timeframe picture walked into a corner. SPX has now spent four consecutive sessions above its weekly band ceiling — the weekly move was fully spent by Monday and everything since has been borrowed range. The monthly ceiling sits at 7761.58, and Friday’s one-day expected range tops out at 7757.76 — the daily ceiling and the monthly ceiling are the same shelf to within four points, with the two-sigma extension at 7805 and the quarterly collar’s short calls at 7890 stacked behind it. Upside into Friday is therefore a wall of coincident supply: a relief pop off a soft jobs number runs into the monthly ceiling inside the first standard deviation, into fresh harvest call supply at every strike on the way, and into the collar shelf above that. Downside is the asymmetry: the abandoned weekly ceiling at 7594 is 1.5% below, with the daily floor at 7662 the only stated level in between — a hot print has an air pocket to work with. QQQ carries a normal box; NVDA’s stall at its weekly two-sigma edge (219.41, tagged to the tick Wednesday, closed under it Thursday) marks the first failure-to-extend of the leaders’ run; DIA’s Wednesday poke above its monthly ceiling reversed — that regime-signal candidate is retracted, not carried. The quarterly bands remain comfortable for every index, which is the frame that says this is late-stage digestion inside an intact quarter, not a top demanding immediate expression.

The magnificent eight against their bands — five names outside, two decision levels

Five instruments enter the weekly expiration outside their weekly expected-move bands — NVDA, META, AVGO, INTC and SPY itself — with IWM at the line and SMH and NFLX within 1.3%. That roster is the mechanical description of what “stretched” means this week: the strike zone the vol-sellers priced last Friday has already been overrun by the squeeze, and everything above it is being carried on borrowed statistical range into a jobs print. Two levels concentrate the whole picture. MSFT 499.81: Microsoft closed on its August monthly band ceiling to the nickel, at the round 500, on day four of a 21-session month — the cleanest pin candidate on the board, with the profit-taking screen showing calls still being written into the breakout. NVDA 219.41: the weekly two-sigma edge it tagged Wednesday and closed 0.19% under Thursday — and traded through in Friday’s pre-market. A Friday close above that line is a genuine two-sigma weekly event, the kind of print that forces a regime question; anything less and the band math says pin-or-fade back toward 210. SPY is pressing its weekly and monthly ceilings simultaneously — the printed weekly two-sigma row on the source board is garbled, and the arithmetic reconstruction of it sits within a dollar of Thursday’s close, so the honest statement is that SPY is somewhere between “at” and “just through” its weekly extreme depending on which number you trust; both are stated, neither silently preferred.

The other half of the eight is the tell for a soft-print morning: AAPL, GOOGL, AMZN and TSLA are all comfortably inside their weekly and monthly bands with 2.6–8.9% of statistical headroom — if Friday bids the tape, the path of least resistance is rotation into the unextended four, not further extension of the stretched five. AMD is the unextended semiconductor (8.8% to its weekly ceiling, 16.4% to its monthly) — of the whole complex it has the most room to run without triggering a band event. MU carries the widest relative bands on the board (a weekly move of 12% is priced, not a signal), MSTR similar at the monthly layer. GOOG has no expected-move coverage on any timeframe — absent from the weekly, monthly and quarterly boards alike — so any GOOG-specific level talk this week is unanchored and this report won’t offer one (GOOGL’s bands are share-class context only). And TLT remains the only instrument anywhere on the boards testing a quarterly band — the floor discussed above, two cents under the close, on the morning of the one print most able to break it.

Friday’s map: the whole board walks in pinned at its ceiling

The forward zone map landed overnight, and its headline is breadth: an unusually large slice of the board enters the jobs-report expiration pinned at the top of its range. Gold at the extreme — GLD closed with 0.21% of headroom to its zone ceiling, the map’s deepest pin, with silver at 88% of its range, the gold miners at 92%, the juniors at 90%, all carrying trend values above the zone (overhead resistance, not support). The index complex sits at 81% of range (SPX/SPY, 1.15% of headroom), small caps at 71% with 1.35% of room, the Dow at 76%, NVDA and AVGO both at 86% with barely 2% of ceiling left. Disney, Pfizer, MetLife, McDonald’s and PayPal closed at or through their zone highs outright. A board this uniformly pressed against its ceilings, walking into a binary macro print on an expiration morning, is the zone-map expression of the same statement the harvest data has made for two days: the upside is a wall of coincident supply; the room is below.

Two structural notes off the same map. The volatility rows are finally quotable, and they are unanimous: VIX closed at 21% of a 13.91–19.86 zone whose trend layer is statistically dead — as are the Nasdaq vol, small-cap vol, and nine-day vol rows — a map-level portrait of the vol-supplied regime, with gold volatility the only vol row on the board carrying a live uptrend. And the traditional defensive floor is broken where the crowd expects it to hold: CVS closed below its zone floor with a reversed trend, UnitedHealth, NextEra and Simon Property sit at their floors with reversed trends of their own, utilities and real estate carry dead ranges. The defensive bid the dark pools showed — Lilly, Procter, Pepsi, Thermo Fisher on expanding volume (block-derived nets, zero intraday prints among them, so the price action carries the verdict) — is going to quality defensives, not bond-proxy defensives. The rotation is real and it is selective.

The bond floor is the cleanest line on Friday’s board

TLT closed at 82.52 — two cents above its quarterly band floor at 82.50, the only quarterly-band test anywhere on the board. And Friday’s one-day expected range for TLT reaches 81.98: a merely normal down day breaks the quarterly floor mechanically. That is the cleanest single line Friday offers, because everything hawkish that accumulated today expresses through it: the June job-openings report showed quits at a near one-year high and the openings-to-unemployed ratio at its highest since January 2025 — labor is not weakening — and the overnight wire carried Fed governor Kevin Warsh reportedly open to a September hike if inflation runs hot. A jobs number hot enough to price that tail pushes TLT through 82.50, and a genuine break there (more than about 0.8 points through, not a two-cent pierce) is the rates limb of the wider risk chain below. Duration buyers who have been accumulating for a week get their answer at the same line.

The AI-leverage names cracked the same day the short thesis went public

Nebius fell 13.3% to 189.88 on $392M of dark-pool distribution the same day reports surfaced that Michael Burry has opened large shorts in NBIS (from 211.77 — already 10% onside) and Oracle (from 144.63; ORCL closed 143.47), framed around off-balance-sheet obligations, uncommenced leases, and purchase commitments in the AI infrastructure build. Set aside the celebrity; our own tape had already graded half his thesis: the profit-taking screen flagged ORCL as a harvest name today — institutions cashing a +12.5% trailing run through the options channel, on a name the forward zone map shows mid-range with 6.5% of headroom under an overhead-red trend value — and CRWV, the most leveraged pure-play in the cluster, fell another 5.1%. Meanwhile GOOGL — down 1.3% on $1.78B of dark distribution — is marketing an investment-grade bond offering in as many as ten parts out to forty years: the off-balance-sheet liabilities Burry is naming are simultaneously migrating on-balance-sheet at about 155bp over Treasuries. The long side of the same coin is real: AI-related private investment ran a record $1.5T annualized in Q2, data-center construction $68B annualized and now $25B larger than office construction. Both things are true — the capex is real and the leverage financing it is real — which is exactly the shape of a theme that stops rewarding everyone at once and starts choosing balance sheets. The tape chose today: the cash-funded names went flat to up, the leverage-adjacent names down 5–13%.

Memory: crowded, two tells red, and the bid underneath still real

The memory complex took its first broad hit — MU −1.3%, DRAM −4.3%, plus SanDisk and Western Digital’s earnings wrecks — while TSM +1.0% and SMH +0.3% held the surface. Two structural warnings now sit on the theme. First, the crowding is objectively extreme: the memory fund DRAM has absorbed roughly $27B of cumulative inflows since an April launch — larger than ARKK’s entire mania peak — with daily volume touching $8B, and the fund already trades 40% below its June high. Flows of that shape have historically marked the crowd’s position, not the smart money’s. Second, TSM’s options channel turned into a call-supply factory today: $100.3M of calls sold on a green day, and the freshly-opened far-out-of-the-money bucket — the genuinely bearish kind of call selling — was the dominant share of it. A green stock whose upside is being sold in size by two different measures is a name being lent against, not chased. The complex’s dark-pool bid is still standing — TSM printed $1.08B of accumulation, the SMH dip keeps getting bought — so the theme is intact but late: crowded at the fund layer, capped at the leader, and now two red tells deep after the failed DRAM roll.

Metals: the dark bid grew while the price stood perfectly still

GLD closed +0.01% — dead flat, pinned 79 cents under its August monthly ceiling at 390.46 — and underneath that stillness the dark-pool tape expanded 89% day-over-day to a +$519M net buy. Price going nowhere while the dark bid nearly doubles is the accumulation signature this framework exists to catch: the ceiling is a shelf someone is loading under, not a wall being rejected. GDX held above its weekly two-sigma band for another session, WPM broke to new ground at +2.0%, and the structural-buyer story got a corporate face: Tether bought 14 tonnes of gold in Q2 — 146 tonnes total, roughly $18.8B, now the largest known private holder outside central banks, out-buying all but four central banks in the first half. With the dollar index below 100, nothing blocks the thesis. The discipline is price-level, not conviction: no adds at the ceiling; the dark bid says dips toward 384 are sponsored; above 390.46 on volume the August range re-opens upward.

Oil reversed on a headline that reloads the yen channel

Crude confirmed its floor reversal: the September contract closed 78.23, holding above the 77.5 line that Wednesday’s channel thesis named as confirmation, after Iran published a restrictive Strait of Hormuz draft — the second session of a +4.6% two-day swing off a floor the zone map had marked with a dead trend layer and 20%+ upside bounds. USO +3.5%, XLE +1.5% off its lows with its own distribution footprint overridden by the tape today. The reason this matters beyond energy: rising oil is the reflation input that pushes Japanese yields higher, strengthens the yen, and pressures every carry-financed position — and the Korea limb of that chain fired in sympathy the same session, EWY −3.0% on $457M of dark distribution. The channel is loading limb by limb: oil confirmed, Korea selling, the bond floor under test. What has not fired: a fast yen move through key levels, a break in the long Japanese bond, the dollar failing at 99.9, and volatility off its floor. Loading, not yet firing — but every limb that could confirm today, did.

Sentiment: the moonshot stalled exactly where they always stall

Thursday’s gauge landed overnight: 68.9 — the first one-day decline since the fear-to-greed moonshot began. The five-day gain is still +26.5 — decelerating from the record +34.5 but still above the velocity threshold, so the squeeze-completed reading stands and the momentum flag is technically still bullish. The level is the message: 66–73 has been the cycle’s exhaustion band, where every prior visit either round-tripped within three sessions or preceded a flush — and the gauge just ticked negative from inside it, one session before a jobs report on an expiration morning. Not a contrarian extreme in either direction (the 80-plus arm never armed); a complacency caution that informs sizing, not direction. The crowd that was fuel on the way up is now long, stalled, and holding its breath into the print.

The dealer book, re-cut overnight: short the jobs week, long the monthly

The stale reading is retired — and the fresh cut says the same thing, smaller and more precisely. Wednesday’s dashboards had dealers short the expiring weekly book to the tune of −$3.45B, and this report’s draft carried that number flagged as one session old. Thursday’s dealer panel, landed overnight and re-cut from the source: dealers enter Friday’s expiration short −$2.59B at the expiry itself — about a quarter smaller than the stale figure, same sign — and the short book does not stop there: every expiry through Wednesday August 13 is dealer-short, roughly −$4.09B cumulative across the front week. The book flips long at August 14 (+$0.81B) and is +$2.35B long at the August 21 monthly (down from +$2.8B Wednesday, sign intact), long again at month-end and out through October. The structure in one sentence: dealers are positioned to amplify whatever happens between Friday and next Wednesday, and to pin whatever remains into the August 21 expiration.

The intraday gamma layer sharpens the same picture. All three index products closed on their maximum-negative same-day gamma strikes — SPY at 768.56 inside a 767–769 pocket where the 769 strike collapsed from mildly positive to deeply negative over the final three hours, SPX at 7,709.96 dead on the 7710 strike (the largest negative on the board), QQQ at 714.65 on the 714 strike — with the stabilizing positive shelves sitting overhead at SPY 770–772, SPX 7740–7760, QQQ 718. SPY closed below the floor of the dealers’ own expected-range box. If Friday’s same-day book rebuilds the usual shape, the jobs print lands on short-gamma-at-spot: the first move gets amplified in either direction until price reaches the shelf above or exits the pocket below. And the daily delta-exposure series flipped red — about −$1.1B — two sessions after printing its five-month record positive, the sharpest negative delta impulse of the whole lookback. One bar is a warning, not a regime; but the rate of change is the signal, and it points the same way as everything else this week.

The premium map and the standing books agree on the week’s shape. Friday’s expiry took the largest fresh two-sided premium purchase on the whole expiration map — roughly +$16M of calls and +$18M of puts bought together — which is event insurance both ways, not a directional wall; the same expiry carries the lightest standing net-premium footprint of the front four (−$31M), meaning jobs-day convexity is genuinely not crowded. Meanwhile the governing August 21 book’s seven-week recovery in standing premium reversed this session (its first down-tick in four sessions — protection being re-added at the monthly), the August 28 line bled to a new low (the post-expiration week quietly accumulating protection), and the single largest row on the entire map sits seventeen months out: roughly $50M of January 2028 premium sold, $41M of it puts — long-dated put underwriting, the income side of this regime writ structural. Read together: amplification window through next Wednesday, pin-and-relief phase after, with the monthly expiration as the hinge — the same shape three independent constructions (dealer inventory, standing premium, and the timing charts below) each drew on their own.

DEALERS Delta book by expiry (re-cut; inventory deltas with matched structures already netted by the panel — positioning, not a directional-flow residue): 08/07 −$2.59B (calls −0.98 / puts −1.61) · 08/10 −0.80 · 08/11 −0.24 · 08/12 −0.32 · 08/13 −0.14 → front-week cum −$4.09B · 08/14 +0.81 · 08/21 +$2.35B (calls +4.47 / puts −2.12) · 08/31 +0.81 · 09/18 +0.69 · 10/16 +0.83. Same-day gamma at the close: SPY 769 strike −$3.0B · SPX 7710 −$5.8B · QQQ 714 −$3.4B; shelves 770–772 / 7740–7760 / 718. Flow Map 08/07: +16M calls / +18M puts bought (two-sided). Flow Timeline: 08/21 line −167M (first down-tick in 4) · 08/28 −144M new low · 08/07 −31M. Jan-2028: ~$50M sold, $41M puts. Market DEX: −$1.1B vs record +$3.5–3.7B on 08/01–08/03.

Friday: the jobs report lands on a book built to amplify it

Friday is the compressed version of everything above: the July jobs report on a weekly expiration, into a market capped at 7757–7761, unhedged underneath, pressed to its zone ceilings, with the hawkish tail priced at 55% for September and keyed to this exact number — and now with the dealer map measured, not assumed. Three shapes, ranked by the mechanics; deliberately unquantified, because expiration mornings suspend normal odds:

• Amplified-open-then-decide (base): the same-day options book rebuilds short-gamma-at-spot — all three products closed on their maximum-negative strikes — over a dealer book short $2.6B at the expiry. The first move off the 8:30 print gets amplified either way: upside stalls into the SPY 770–772 shelf under the 7757–7761 double ceiling; downside runs an air pocket to 7662 with nothing structural between. The weekly expiries — including the SpaceX put floor — burn at the close. The timing chart independently projects this session as a fade.
• Hot print (the tail — trimmed, not retired): rates reprice the September tail, TLT breaks the quarterly floor (a normal down day reaches it mechanically), and the dealer-short front week amplifies through 7662 toward 7594. Honest update: the short book is a quarter smaller than Wednesday’s stale reading, and the one credible mechanism for the floor holding — the Treasury’s bill-heavy issuance tilt — is on the record. Watch the wages component, not the headline; composition is the trigger.
• Soft print (the squeeze the overnight tape is front-running): the hike tail un-prices, software gaps hold, and the relief pop tests 7757–7761 with the after-hours cohort as fuel. Through it on volume: 7805, then the collar’s short calls at 7890. The trap in this branch has a name and a calendar: the Friday-catalyst fakeout — a strong close on the news reversed Monday — and a book that flips dealer-long only on August 14. A Friday rip does not repeal the amplification window that follows it.

The timing charts: two-for-two, and Friday is the projected fade

The August projection has hit both of its inflections so far — the trough it placed at the turn of the month printed with the July 30 low and the August 3 launch, and the local top it placed at August 4–5 printed with Wednesday’s squeeze high, exact. Its next call is the one Friday tests: a fade completing around August 7, a minor bounce into Monday the 10th, then a decline into the month’s key buyable low in the August 14–17 window, followed by the strongest rally of the month into the 25th — a leg that would run straight through the August 21 expiration where the dealer book is long and pinning. The inverse orientation of the same chart missed its first leg by all three sessions and is dead for the month. Discipline note, always: these charts give dates and shapes, never sizes — every magnitude in this report comes from the expected-move bands, not from a projection axis. The chart’s value tonight is the agreement: its weak window (now through mid-month) is the same window the dealer inventory is short, and its rally window is the same window the dealer inventory is long. When the clock and the positioning say the same thing from independent sources, the shape earns weight.

The bond chart’s moment is now: its final-low window for Treasuries runs roughly August 8–12 — opening today — with TLT already sitting on the quarterly floor the price map independently flagged. Hold-and-turn inside that window selects the chart’s bullish orientation (and validates Dale’s duration-scarcity mechanism); a decisive break of the floor selects the inverted orientation, in which bonds don’t bottom until September (the room’s consensus). Until TLT answers, no bond-direction claim gets to cite these charts — the floor is the referendum.

The commentary room: one bond diagnosis, four different endings

Six of seven macro voices overnight are effectively short duration into the jobs print — the bond vigilantes are awake (Hanke), the Treasury’s own refunding documents show a $1.37 trillion borrowing wall in the back half (ClearValue), the 30-year’s weekly breakout is holding (Trading Apologist, who also has September hike odds at 55% and keyed explicitly to tomorrow’s number), and the yen defense is being read as dollar sacrifice (Davinci). When every voice in the room agrees the bond market is in charge, the crowded trade is short duration into a payrolls print — which makes the TLT bounce off the 82.50 quarterly floor the pain trade, and exactly one analyst supplies a mechanism for it: Darius Dale’s reading of the refunding’s bill-heavy tilt (61% of Q3 issuance in bills vs a ~41% norm) as manufactured duration scarcity, plus a widened Fed facility that lets Japan defend the yen without selling Treasuries. His is the only framework in the room in which the floor holds — worth exactly that weight, on the one line everything else keys off. His composition note matters more than any level: a hot payrolls headline with soft wages is not a hike trigger; hot wages is. Watch the internals, not the print.

On equities the same room splits four ways from one diagnosis — bubble-pops-on-rates (Hanke, September–November window), bubble-inflates-on-the-Treasury-bridge (Dale, twelve-month bull), Fed-prints-hard-assets-win (ClearValue), and buy-defined-pullbacks-but-institutions-are-selling-closes (Trading Apologist, whose five-straight-days-of-close-selling observation independently confirms this desk’s monetization read). The trading room adds the texture: nobody argues for unimpeded upside — the only unreserved longs are mechanical systems whose own operators published cautions against themselves; the AI-infrastructure second tier (Nebius, Oracle, CoreWeave, SanDisk) is now the consensus short across the celebrity and newsletter layers, which is a crowding warning on this desk’s own lean — late entries there are squeeze fuel; and one commentator named the exact mechanic this session should fear most: the record-adjacent Friday-catalyst fakeout — a strong Friday close on the news, reversed Monday. On the covered session, that is not a hypothetical; it is scenario three with a calendar attached. The practitioner segment on hedging restated this desk’s vol discipline from the other side of the table — long puts are the worst hedge because the skew makes you pay up for them, VIX calls failed their holders in the August 2024 spike, and the right structure is short calls and call spreads, sized first — which is, line for line, the book the operator already runs.

Unusual structures

The SanDisk floor-write: $299M of puts sold into a −6.8% hole

Covered in full above — the size, the strikes below spot, the fresh-open prints above existing interest, and the premium collected at implied vols of 124–154 against back months near 100–107. The one addition for the structures ledger: the largest single prints in the complex were deep-in-the-money puts sold at the bid in the same window, the stock-substitute form of the same bet. Whoever this was legged into synthetic long exposure at the bottom of the range through three instruments at once.

SpaceX carries a genuine put floor into its expiration — and it expires tomorrow

SPCX printed +6.1% on $2.36B of dark accumulation, the second most active options line in the entire market. At tomorrow’s expiry the 100 strike holds put open interest at 14.6x its call interest — a genuine one-sided hedge floor, not two legs of a financing box — with a second floor at 110 at 3.2x. Real floors, directly under a squeeze, with the dark tape’s own $684M shelf at 114.90. The mechanical point: all of it expires at Friday’s close. Whatever is holding this squeeze together structurally is a one-day asset.

TSM: two independent measures say call supply is building on a green stock

$100.3M of TSM calls sold on a +1.0% day — a residue that survived every structure filter — and separately, the moneyness screen graded TSM the day’s clearest case of bearish-opening call sales: far out-of-the-money, freshly opened, in size. When the sold-call channel and the moneyness channel agree against a rising price, the read is a name being capped by someone with size and patience. It cannot veto TSM’s uptrend — the price won the day — but it is the same signature the index complex carries, expressed in the semis leader.

NXPI: the only pure bearish-opening print in the file

NXP Semiconductors closed +0.3%, quiet tape — and 100% of its sold-call premium graded as far out-of-the-money, freshly opened, above existing interest. One name, one clean signature, no harvest component at all. Small in dollars, but it is the only symbol in the file where the call selling has no innocent explanation. Worth a line on the watchlist into next week.

The defense tape went dark — literally

RTX, BA and LMT printed zero darkpool prints among them — every dark dollar on all three names crossed as blocks, which carry no directional information. On a day BA fell 3.3% while RTX and LMT rose, the entire institutional layer of the defense complex is unreadable. Flagged not as a signal but as an honesty note: any confident read of Thursday’s defense flow would be invented.

A commentator’s five unusual trades, taken to the actual prints

The most-watched flow commentary of the evening walked through five “unusual” institutional trades; this desk pulled every one from the raw tape and decomposed it. Two survive fully: the metals-and-mining ETF call package (XME September 117/130, ~$794K — real, bullish, and he missed the larger one: an October 120/140 package at twice the size printed that morning) and the GLD September 410/430 spread ($20.4M collected — correctly read as a ceiling-through-September call, not a fade-gold call). Two survive partially: the Whirlpool “puts bought in two expirations” is actually a roll — the September legs were sold to close at size matching open interest and only the November legs are new, a timing concession that cuts the citable bearish flow to about half a million dollars; and the SanDisk “$8M bear call sale at 1430” is half of a $54M short strangle whose put side collected twice the call side — the same trader sold nearly $35M of premium betting SanDisk does NOT fall below ~1350, which inverts the bearish gloss entirely. One dissolves on contact: the “insane insider call buying” in First Solar decomposes into calendar spreads printed at mid with no readable side, on a name whose fifteen-day dark-pool footprint just made a new low — whatever front-ran the tariff headline, it is not visible as institutional accumulation in either channel. The meta-lesson is the same one this framework paid tuition for in July: single-leg reads on multi-leg structures manufacture direction that is not there — and on the very names he graded bearish, the measured signature of the day was premium selling into panic, the opposite of institutional fear.

The rest of the mega-cap board, in one pass

AAPL +0.5% to 312.41 — quiet accumulation on the census read, but the dark-pool label net ran heavily negative on thick tape, so the two channels disagree; with Friday’s band 308–316, it is a range name until one channel wins. META +0.2% to 589.90 with a genuine wrinkle: its fifteen-day distribution footprint met an up-tape today — the first possible-turn signal in two weeks — while its options channel still leaned to call selling; 578.90 is the Friday floor that keeps the repair alive. GOOGL −1.3% to 357.75 and GOOG −1.0% to 356.62 remain the only mega-caps being genuinely sold — harvest-graded again, now with a 40-year bond deal on the tape; 352.33 is Friday’s line under GOOGL, and below it the digestion story becomes a distribution story. TSLA −0.6% to 319.53 printed a small structure-clean bullish residue against a red tape — a disagreement that supports patience, nothing more; 312.98–326.08 Friday. INTC −1.2% back under $100 to 99.81 — the round number is the tell; its recent run is consolidation per the profit-taking screen, not exit. NFLX −0.7% to 73.69, drifting mid-band, no institutional footprint worth a sentence. MSTR −1.6% to 96.85 with the crypto complex soft (IBIT −0.7%, COIN −3.0%) — the covered-call anchor near its monthly ceiling around 131 stays the sleeve’s working plan, and nothing in today’s tape touched it. MU −1.3% to 881.47 and AMD +1.5% to 489.28 are covered in the memory and book sections above; AVGO 420.57 holds the cleanest accumulation ladder on the semis board with Friday room to 430.34.

The operator’s book: torn down at the open, rebuilt tighter by the close

Thursday was a full reset, executed in two passes. The morning session closed roughly forty-five structures — every AMD short-call spread (the September 580/600 near flat, the August 470/475 at a small loss — that thesis closes its ledger entry: exited, not run over), the MSTR August 130 covered call (bought back at a third of the credit), the entire HUT complex including the short shares, the Intel spread from the “highest-conviction” lesson (small loss, time was winning but the reset won faster), the Microsoft, Tesla, Oracle, ServiceNow and smaller short-call structures, the Palantir December diagonal’s short leg — and, notably, the hedges too: the September QQQ put spreads and the inverse-Nasdaq sleeve went out with everything else. The afternoon rebuilt the same thesis at nearer dates and tighter strikes: some twenty-five fresh defined-risk structures concentrated at the August 21 monthly and September 18 — NVDA 225/230, CRM 200/210 in both months, IWM 305/310 twice, ADBE 275/285, AMGN 420/430, PLTR, HOOD, MSTR, CRWV, SPCX, NBIS 220/240, LEU, WMT and the small book — plus the re-issued IBIT covered calls (August 21, 37.5 strike, moved off the expiring weekly), and the two put positions the operator half-remembered opening: QQQ August 14 710/705 — dated, knowingly or not, to the exact session where the dealer book flips from amplifier to pin and where the timing chart puts the month’s low — and EWY September 160/150, the Korea expression of the oil-yen-carry chain. The margin loan came down another $7.8K in two sessions. The book that emerged is the same harvest thesis with less clutter: short fresh call supply into the August 21 lid, hedged into the mid-month window, defined-risk everywhere.

Two reconciliation flags, so nothing rides on memory: the SPY September 650 put recorded earlier this week is not in the account and no Thursday order closed it — treat as closed before Thursday or mis-recorded, either way off the books. And the ARM shares that carried the August 350 covered call are no longer in the retirement account export, with no ARM options anywhere — that construction should be graded as closed-or-moved, not as a live overwrite. Both flagged to the operator; neither changes a current position.

Triage: the short book against the overnight tape

The operator’s worry, stated plainly: the after-hours software pops and the overnight drift look like the start of a squeeze over a freshly-rebuilt bearish book. The triage below is the worry converted into numbers — per position, against its strike, its band, and its own pre-registered exit line. Overnight marks as of roughly 5:00 AM ET (pre-market, unofficial, thin): SPY 769.9 (+0.2%), QQQ 718.5 (+0.5%) — sitting exactly on its overhead gamma shelf — IWM 298.8, NVDA 220.0 (through the 219.41 weekly two-sigma edge), CRM ~190 (+1.7%), ADBE 262 (+0.7%), PLTR 157.5 (+1.0%), HOOD 91.6, MSTR 97.6, CRWV 87.2 (+2.2%), SPCX 116.8 (+1.7%), NBIS 197.5 (+4.0%), SNDK 1,304.7 (+3.7%), TEAM 144.3 (+31%), NET 326.6 (+15%). EWY 165.9 (+1.1% — against the new put spread), TLT 82.46 (holding its floor by pennies until 8:30).

Genuinely at risk today

CRM is at its own kill line before the market even opens. The exit condition written into the ledger when the spread was opened — “dies on a reclaim of 195 / close above 193” — sits 1.5% above the overnight quote, with Atlassian +31% and Cloudflare +15% gapping the entire software complex. The 200 short strike is still 5.3% away and the September legs have time, but the discipline is the discipline: a Friday close above 193 exits the CRM spreads by the rule set at entry, no renegotiation on the day. PLTR is the book’s tightest strike and its least framework-aligned position. The short 160 call sits 1.6% above the overnight quote, on a name whose zone trend is an intact uptrend with 6% of headroom to 165, and whose own rollover trigger — the 200-day line at 152.44 — has never broken. This is the one position that genuinely fits the operator’s “I shorted a consolidation” self-diagnosis: the framework’s stated discipline was no bearish anticipation before the trigger, and the strike was sold above an unbroken uptrend anyway. Mechanical management, borrowed from the Intel lesson: out on a close over 160, or when the short call’s delta crosses ~0.35 — whichever comes first. The 170/175 spread above it is the backstop, not an excuse to hold the 160.

Pressured but defensible

SPCX (+1.7% overnight, short 120 strike 2.7% away): the unlock squeeze is real, but its structural support — the genuine 14.6x put floor at 100 and 3.2x at 110 — expires at today’s close. Decision point is Monday’s naked tape, not today’s squeeze; surviving the day is the assignment. HOOD (91.6 vs short 95s, 3.7%): distribution footprint intact at the fifteen-day layer but the three-day pace has been easing; watch, don’t add. IWM (298.8 vs short 305, 2.1%): pure payrolls beta — the zone ceiling at 302.28 is the first defense, and a soft print that runs small caps through it threatens the August legs; the September 305/310 has room to breathe. ADBE (262 vs short 275, 5.0%): the software gap lifts it by sympathy, but 275 sits at the far edge of its zone; reassess above 270, not before. EWY +1.1% against the new put spread — the yen channel is loading, not firing, and this position was sized as a September thesis, not a Friday trade; no action on an overnight drift.

Structurally protected

NVDA short 225/230: the short strike sits above both the weekly two-sigma edge (219.41) and the monthly band ceiling (224.19). The overnight print at 220 is the two-sigma line itself — for the 225 strike to be threatened, NVDA has to close through a weekly two-sigma and approach a monthly band breach in the same move, a statistical regime event, not a drift. The position is short exactly the zone the vol-supply complex is short. MSTR (needs +18% to the 115 strike), CRWV (+15% to 100), NBIS (+11% to 220 even after its +4% bounce — and the bounce is happening on a tape whose measured flow was harvesting the panic, not chasing the recovery), AMGN (red overnight, 420 strike 4% up and the zone ceiling in between), WMT (fading, 115 nearly 3% up on a defensive that just broke lower), plus the small book — LEU, SMCI, GFS, SYM, TE, NNE — all carry strikes at or beyond their names’ statistical ceilings. These are the positions the harvest tape itself endorses: short fresh call supply at levels institutions are also selling.

The mitigants the fear forgets

Every position is defined-risk — the maximum damage is known and paid for. The book is not one-directional: the QQQ 717/720 call spread bought Thursday is long the exact shelf the index is gapping toward, the VIX September calls and NVDA September puts were kept through the reset, and the new QQQ August 14 put spread is dated to the precise window where the dealer book flips from amplifier to pin — and where the timing chart puts the month’s buyable low. The thesis was never “Friday breaks.” The thesis is capped-upside into August 21 with a weak window through next Wednesday — and nothing that happened overnight touched that structure. Surviving the morning print with the tight strikes managed is the whole job; the book’s window opens Monday.

The seasonality complaint, answered with data

“The one year I build a bearish August book, August refuses to be bearish” — three answers from the file. First, the August projection this desk actually tracks was never linearly bearish: its shape is trough (hit, July 31), top (hit, August 5), a shallow fade into Friday, a mid-month low around August 14–17, then the month’s strongest rally into the 25th. A bearish-tilted book with mid-August dates is aligned with that shape — the error would be needing the whole month to bleed. Second, the seasonal-bear trade is objectively crowded this year — a 34-name public short book, celebrity shorts in the same AI-infrastructure names our gates flagged, a commentary room unanimously short duration — and crowded seasonal trades get squeezed before they get paid; the squeeze IS the entry mechanism, not the refutation. Third, the harvest tape argues August resolves capped, not crashed: 81% of call selling is still profit-taking, protection is being re-added at the monthly, and the market is pinned under a triple ceiling. A capped market is a credit-spread market — which is the instrument this book already holds. The direction of the frustration is understandable; the instrument choice was right.

Am I too early to be bearish? The reconciliation in one paragraph: the greed gauge (68.9, first down-tick of the entire moonshot) measures the crowd; the flow measures the institutions — and the institutions spent the week doing exactly what this book does: selling calls against winners (81% of all call premium sold), rotating dark-pool participation out of mega-cap tech and into quality defensives and gold, flipping the daily delta impulse negative, and re-adding protection at the monthly expiration. Defensive positioning is not early — it is the institutional consensus expressed at retail scale. What would be early is needing the index to break this week: no signal count licenses that, the sentiment velocity is still technically bullish, and the after-hours tape just proved demand exists in washed-out names. Capped-upside expressions above the shelves: aligned. Breakdown bets before the triggers (TLT through 81.7, a hot-wages print, the yen channel actually firing): early by the framework’s own rules.

Why the bearish reads feel off-sides — the clock, not the eyes

Every bearish input the operator names is real and measured. Yields pushing higher: the long bond is sitting on a quarterly statistical floor with the 30-year’s weekly breakout holding — confirmed. Oil rebounding and building: two sessions, +4.6%, holding above the 77.5 confirmation line — confirmed. Mega caps distributed: the profit-taking screen graded MSFT, GOOGL, AMZN, NVDA, PLTR and ORCL as harvest names this week, AAPL printed the board’s largest negative dark net on thick tape, the index vehicles are in their sixth session of dark-pool outflow, and the daily delta impulse just flipped negative off a record — confirmed, confirmed, confirmed. The reads are not wrong. What is wrong is the expectation that they price immediately in a tape where volatility is supplied and dealers are paid to pin. In this regime, bearish inputs do not transmit to the index on observation — they transmit when a mechanical window opens. The window is measurable and it is dated: dealers are short every expiry through next Wednesday, the projection’s buyable low sits August 14–17, and the written call supply that caps the market expires August 21. The inputs are fuel; the window is the match.

“There are no bears left” is an index-lens illusion. The bear market is already running — name by name, under a pinned index: AppLovin −20%, Datadog −19% on a beat, Celsius −18%, Axon −14%, Western Digital −13% on a beat, Nebius −13%, The Trade Desk −24% overnight. That is what a distribution regime looks like when the index itself is the cap: the invoice routes around it. A book of single-name call credit spreads is short exactly where the bear actually exists — the index-level scoreboard just refuses to show the game. And the “bulls’ only two cards” — semis already corrected, VIX low — are not really the bulls’ case. The measurable bull case is narrower and dated: a dealer book that flips long August 14 and pins +$2.35B into the monthly, the projection’s strongest-leg window August 17–25, sentiment velocity still technically positive, and after-hours proof that washed-out names still catch bids. Which yields the honest month-shape answer to “do the bulls win August?”: the front of the month through roughly the 13th–17th belongs to the bears’ window; the back half, into and past the August 21 expiration, structurally favors the bulls’ pin-and-relief. The operator’s fear has the sequencing inverted — the bears are not gone until next month; their window opens next week, and it closes at the monthly. As for the low VIX: the forward map shows the entire volatility complex — VIX, VXN, RVX, VIX9D, VVIX — dead-ranged; that is not a bull argument, it is the mechanism that grinds both directions, crushes every spike, and makes premium selling the only expression that has been getting paid. It is the argument for the book the operator already holds, and against the despair about it.

Portfolio names on the wire

Across the standing roster, healthcare led outright — BSX +3.3% on $157M of dark accumulation, NVO +3.2% — while the rate-sensitive defensives took the day’s damage: HON −3.0% and homebuilder DHI −3.6%, both on distribution tapes, both the logical casualties of a session where the long bond sat down on its quarterly floor. The defense trio split — BA −3.3% while RTX and LMT rose — and none of the three is pinned: the forward zone map gives BA and RTX about 6% of headroom each and LMT 9%, all mid-zone with intact uptrend layers, which leaves them room to absorb a payrolls morning either way. The nuclear sleeve drifted red (CEG −1.5%, BWXT −0.9%, CCJ −0.7%) with no institutional footprint worth naming. WPM +2.0% was the metals sleeve’s standout. AVGO +0.6% remains the most internally consistent accumulation story among the semis holdings, with one caveat logged: Thursday printed its first negative dark-pool label day of the run, on sharply lower volume — noted, not acted on. And the profit-taking screen’s harvest grades on AMZN and GOOGL landed the same day Jeff Bezos filed a $346.5M share sale — the insider channel and the options channel telling the same story about where in the cycle these names are.

What landed overnight, and what it changed

The draft of this report shipped against a deliberately incomplete evening; everything missing has now landed and is folded in above. What changed: the dealer re-cut retired the stale short-the-weekly reading and replaced it with something sharper — short the whole front week, long the monthly — which upgraded scenario one from “pin” to “amplify-first-pin-later” and dated the book’s windows. The forward zone map turned “the index is capped” into “the whole board is pinned at its ceilings, with the vol complex (VIX, VXN, RVX, VIX9D) dead-ranged and the bond-proxy defensives broken.” The sentiment gauge printed 68.9 — the first one-day decline of the entire fear-to-greed moonshot, from inside the band where these excursions historically stall. The commentary layer added a crowding warning on the consensus shorts and one credible counter-case on the bond floor. And the after-hours earnings tape added the countersignal — the biggest beat-paid software cohort of the week — that Friday now grades. What did not change: the harvest verdict, the absorption failure, the band arithmetic, and the two-speed structure. Those were closed facts of Thursday’s tape and remain the spine of the read.

BOTTOM LINE

The harvest regime is two days old, already collecting, and now fully mapped. A market that spent two weeks selling its own upside and buying no protection watched clean earnings beats get sold 13–19% for a third straight session — while the index never moved, because the index is the cap. The overnight additions completed the picture: dealers are short every expiry through next Wednesday and long $2.35B at the August 21 monthly — amplify first, pin later — the whole board enters Friday pressed against its zone ceilings, five mega-caps sit outside their weekly bands, and the crowd’s greed gauge just printed its first decline from inside the exhaustion zone. Friday compresses it into one morning: a jobs report (watch wages, not the headline) on a weekly expiration, over a bond two cents above its quarterly floor, with an after-hours software cohort gapping up into the sellers’ shelf and a same-day gamma book built to amplify the first move. The asymmetry is stated, not predicted: soft print meets a wall of freshly-written supply at 7757–7761 with a fakeout mechanic waiting Monday; hot print meets an air pocket to 7662 with an amplifying dealer book. The bearish window this book was built for opens next week — through the August 14–17 trough — and closes at the August 21 expiration where the written supply dies. Trade the lines: 7662 / 7594 below, 7757–7761 / 7805 / 7890 above, TLT 82.50 as the macro tell, NVDA 219.41 and MSFT 500 as the single-name referendums, WDC 440/430 and SNDK 1214.80 as the floor-writers’ tests.

Top Trades to Follow

BULLISH · FLOW SNDK — the $299M put-write complex, weekly 1195–1265 strikes sold at IV 124–154 into the −6.8% wreck, several prints above open interest, plus deep-in-the-money puts sold at the bid in the same window (synthetic long entry). The graded test: the 1214.80 dark shelf holds Friday. Follow the shelf, not the writers’ nerve.

BULLISH · DIP-MAP WDC — the +$48.6M dark buy at 457.98 at 3:09pm, placed inside the pre-named 430–460 zone after a 16% intraday range. Executes on Friday stability above 440; dead on a gap through 430. The beat-sold-13% context means this is a trade, not a thesis.

BEARISH · OPENING TSM — $100.3M of calls sold on a green day — a structure-stripped directional residue, matched legs and delta-one already removed — PLUS the file’s clearest far-OTM fresh-opening call-sale grade. Two independent screens, one signature: size is capping the semis leader. Expressed as: do not chase TSM strength; the writers are selling it to you.

BEARISH · OPENING NXPI — 100% of sold-call premium freshly opened far out-of-the-money, the only name in the file with no harvest component to its call selling. Small, clean, and the kind of print that precedes news. Watchlist entry, not a position.

HEDGE · EXPIRING SPCX — the 14.6x put floor at 100 (and 3.2x at 110) under the unlock squeeze is genuine protection, and all of it expires at Friday’s close. The follow: what the squeeze does Monday, naked, without its floor. Any Friday-afternoon put roll to next week is the tell that size intends to stay.

HEDGE · MACRO TLT — the quarterly floor at 82.50, tested to two cents, with Friday’s normal range reaching below it. A close under ~81.7 is the rates limb of the oil/yen/carry chain firing — the graded trigger for scaling the EWY short and yen-strength expressions from watch to position.

HEDGE · CHEAP-TAIL HYG — the September 79 put at roughly $0.40, priced at 0.86x its own at-the-money volatility — the only tail in the entire file priced below its body. Credit-event convexity into the September window at a discount the skew almost never offers. The one place on the surface where buying protection is not paying up.

BEARISH · OVERWRITE GLD — the September 430 call (57K of open interest built in the chase) as the overwrite strike against gold exposure: the monthly ceiling at 390.46 with 0.21% of zone headroom, the options tape already selling calls and buying puts while the dark bid grows — hedged-not-chased is the institutional posture; renting out the 430 line matches it. Dies on a volume close through 390.46.

HEDGE · WINDOW QQQ — the August 14 710/705 put spread (the operator already holds it): dated to the exact session the dealer book flips from amplifier to pin, covering the projected August 14–17 trough window, at the expiry where genuine put shelves (not financing boxes) sit at 700 below. The calendar is the thesis; convexity owned inside the window, retired outside it.

SOURCES

Data date 2026-08-06 (Thursday), covering Friday 2026-08-07 — weekly expiration + July payrolls. FINAL EDITION — all draft-time gaps healed overnight. Inputs: Live Options Flow 0806 CSV (five-stage decomposition, 1,221 symbols, $13.70B gross); Darkpool Market Summary 0806 CSV (3,368 symbols bucketed); monetization ledger (file verdict HARVEST 81.3% / bearish-opening 7.4% / trail +5.1%); structure decomposition on 15 cited symbols (pass); options dashboard 0806 PDF — all 22 pages panel-read (Market Net Flow, same-day flow + gamma by strike for SPY/SPX/QQQ, Market DEX, Flow Map, Flow Timeline, Dealers Diary re-cut, Top Flow, call/put chains, vol-change, cheapies/LEAPs/OTM-OI, sector flow); darkpool dashboard 0806 PDF — all 9 pages (header stats, live trades, sectors, ticker dashboards; bubble panel conventionally skipped); forward zone set for 08/07 — all four artifacts (zone document, zone visual, zones map, range & trend) fully transcribed with colour-layer reconciliation; FOM sentiment gauge 0806 (68.9 GREED, −0.9 / +26.5); forward daily expected-moves board 08/07 (17 single-name rows, closes verified against census); weekly 0803–0807 / August monthly / Q3 quarterly boards + JPM collar (validated transcriptions; quarantined weekly rows excluded); Savino August 2026 projection + inverse + ZB_F bond forecast (timing/direction/shape only); wl1 census 2026-08-06 (632 per-ticker reports; 90+ opened and quoted individually incl. DDOG, TTD, XME, FSLR, WHR, NBIS, CELH, OSCR, U, TEAM, ABNB, LLY, CAT); rolling tracker v63; sentiment tracker (8/6 row appended); thesis ledger (45 open entries graded); market commentary 0806 evening layer — 16 polished transcripts read in full (the flow-commentary deep-dive decomposed print-by-print against the raw CSV; macro digest: Hanke, ClearValue, Darius Dale x3, Trading Apologist, Davinci; trading digest: Tim Knight’s 34-short roster, Conquer, tastylive x2, Mike Jones/Burry, Saylor, DataDash, Rob’s Child); Robinhood main-account reconciliation (80 Thursday orders, 177 open legs strike-resolved, portfolio marks) + Fidelity IRA export 08/07; Robinhood earnings calendar (325 events) + after-hours press sources for unofficial AH moves (census prices authoritative throughout); overnight pre-market quotes ~5:00 AM ET (unofficial, labeled). Remaining gaps: census absent for FXY and VIX (VIX carried from the zone layer); MU/MSTR/SPCX/SNDK/WDC/AXON absent from the zone universe; market-sentiment/intraday-target panel absent from the dashboard capture; prior-session options CSVs unavailable (multi-session flow claims in commentary untestable). Full audit trail: comprehensive_analysis_0806.md.