Daily Report — 07/24/26 · “The Wall Above Your Head”

Scorecard — Grading Thursday’s Report

“The Washout and the Warsh Tail” earns a C+ — the index and cohort work held, two single-name calls were wrong, and both were wrong for the same reason. The report gave Friday as a genuinely two-sided day inside a 735-to-748 band with the 737 shelf as the hinge. The band held, the shelf held, and the session closed between the two markers it named. It told you to treat the Nasdaq’s 688-690 zone as a starter entry only and to hold full deployment until after the Fed — the Qs went straight through it, and the sizing discipline is the only reason that read cost nothing. It kept defense and the build-out receivers on their promotion, and they printed a second green session against a red Nasdaq. It kept the metals block on and the credit read at “leaning, not cracking.” All of that was right.

Two calls were outright wrong. The report named the legacy software complex as the place to express a short, with Salesforce as the flow-pressed axis — and Salesforce closed +4.3% on Friday while the software group led the entire tape. And it kept Korea on a watch with an invalidation level a few percent below spot; that level was taken out in a single session by more than 4%. The honest diagnosis is one sentence: a deep-in-the-money option print tells you delta, never conviction, and that is true whether the position is long or short. The framework applies that discipline to bullish claims religiously and applied it with a blind spot to the bearish one. The Korea stop was the second version of the same error — a level placed inside one ordinary session of that name’s own volatility is not a stop, it is a coin flip.

The one call worth keeping on the record: the report explicitly excluded ServiceNow from the software short, on the grounds that an opening synthetic long had printed into Thursday’s drop and its institutional accumulation ladder was making new highs. It closed +7.4% on Friday, the best large-cap software print of the week.

SCORECARD: Friday distribution called Bear 40 / Bull 35 / Pin 25, range 735-748 — outcome PIN, SPY 738.93, session 737.5-743.5. 737 shelf HELD all session. QQQ starter-only: CORRECT, price went through to 684.23. NOW excluded from the short: +7.44%, best call of the cycle. CRM as short axis: +4.29%, WRONG. EWY invalidation: broke by 4%+, WRONG. Defense 2nd session: RTX +1.74%, LMT +2.46%, GD +1.30%, CORRECT. Net 6 right / 2 wrong / 3 partial = C+.

The Read — A Factor Unwound, Not a Market

Friday was not a risk-off session. It was the most crowded factor in the market being de-grossed while everything else was quietly bought. The arithmetic settles the argument before any interpretation is needed. The Dow finished up. Financials, communication services, healthcare, discretionary, industrials, energy and utilities all closed green. Across the 498 names the pipeline tracks, 42% closed price-bullish — not a number you get on a day the market is being liquidated. And the largest single positive institutional flow of the entire session went into financials, with the technology sector taking the mirror-image negative.

What fell was one cohort with one common property: it is the capital-expenditure supply chain of the AI build-out. Memory. Storage. Foundry. Semiconductor capital equipment. The Korean market that houses the memory duopoly. The neoclouds. The power names. And the credit that finances all of it. That is not “tech.” Apple closed +3.5% on the same tape. Software closed green. Alphabet stabilized on day three. Capital-light beat capital-heavy by roughly 11 percentage points inside a single session, and if you netted the two halves at the index level you got approximately zero, which is exactly what the S&P printed.

The mechanism is positioning, not demand. Nothing in this week’s data says AI demand deteriorated — the weekend headlines said the opposite in size, with a half-trillion-dollar memory alliance and a report of a quarter-trillion-dollar financing arrangement for a single data-center campus. What changed is that the most crowded factor in the market ran into the only two things that reliably force de-grossing at once: a bond market printing weekly closing breakouts across the entire curve, and four capital-spending disclosures compressed into forty-eight hours. When the leadership stops keeping pace, the dealers who are long stock against short calls in those names have to sell that stock, and the selling accelerates into expirations. That is a description of Friday, not a forecast.

TAPE: Sector institutional net: Financials ≈+$9.8B (largest positive), Technology ≈-$8.7B (largest negative), Consumer Cyclical +$2.0B, Utilities +$1.3B, Comm Services -$2.8B. Breadth 208 up / 256 down / 34 flat of 498. Ladder census 68 rising-accumulation vs 73 falling-distribution — near parity on a day the Qs lost a percent.

Your Question, Answered Straight — Is This Week Going to Clap the Late Put Buyers?

The mechanical setup for exactly that is real, it is unusually specific, and it has a shelf life of about four sessions. Here is the case, laid out so you can judge it rather than take it.

• The crowd has disclosed itself. One flow desk described the week’s activity as “a lot of S&P 500 and QQQ put flow — millions upon millions, out of the money, relatively short dated” and called it “super unusual” against a summer that had been persistently bullish. A second desk, reading its own client book, said customers “got way into short-the-pop mode on Friday.” When two independent vendors describe the same positioning without coordinating, that is not narrative — that is the position.
• The protection is now above the market. Every genuinely one-sided Nasdaq put strike expiring this Friday — 690, 695 and 700 — sits above a 684 spot. Those contracts are in the money. The dealers who sold them are maximally short-hedged, and the negative-gamma amplification that dragged price down through those strikes is spent. What is left is a hedge book that must be bought back on any move up. The S&P’s own 740 strike is in the same condition, one dollar overhead.
• Dealer gamma is negative above spot, not below it. On Friday the only meaningful positive-gamma shelf on the S&P tracker sat at 737, below the close; 739, 740, 741, 742 and 745 were all deeply negative, with 740 the most negative strike on the board. Negative gamma above spot means dealer hedging amplifies an upward move as readily as a downward one. That is the same structure a hedging-mechanics desk flagged going into Thursday’s decline — and it cuts both ways by construction.
• The mood is already washed out. The sentiment gauge printed 33.8, deep in fear, and rose only fractionally on a flat index day. Technology breadth — the share of tech names on point-and-figure buy signals — collapsed into the high twenties with its momentum oscillator in the mid-thirties, a zone that has marked washout in every prior instance on the chart. There is no euphoria to unwind here.
• And the calendar supplies the trigger. A Federal Reserve decision Wednesday afternoon, then Microsoft and Meta Wednesday night, then Apple and Amazon Thursday night. Four of the largest weights in the index resolve their single biggest uncertainty inside two sessions.

Now the honest counterweight, because a squeeze case with no counterweight is a sales pitch. The put buying in the AI build-out names was not crowd noise — it was institutional, it opened brand-new positions, and it was priced at ordinary at-the-money volatility rather than panic volatility, which is what genuine hedging looks like rather than lottery-ticket chasing. And it is dated the seventh of August. Meanwhile the bond market printed weekly closing breakouts on the two-year, ten-year and thirty-year simultaneously — the thirty-year’s first weekly close outside a range it has held since October of 2023. That is a structural event, not a level event, and it does not un-happen because oil fell two percent.

The synthesis: this is the squeeze week and August is the risk window. Not because that is a tidy story, but because that is where the money actually sits — near-dated index protection flat, single-name protection concentrated in one sleeve and dated the seventh, far-dated index downside written in size, and an in-the-money put wall directly overhead. The way to be wrong is to treat Monday’s gap as the resolution of anything. The way to be right is to demand that the gap hold, and to trade the calendar the flow actually bought.

TAPE: Genuinely one-sided put open interest expiring 7/31 (call/put ratios, box-parity strikes excluded as financing): QQQ 690 19.8x · 695 18.5x · 700 8.5x — approx 142,000 contracts, all above a 684.23 spot. SPY 740 9.2x (72,202 puts), 720 15.0x. Sentiment 33.8 fear (+2.8). Tech bullish-percent 29.58, oscillator 36.95.

The Discipline Section — The Index Book Is Bullish, and It Is Dated 2027

This is the part most readers will get wrong this week, so it goes near the top rather than buried. Strip Friday’s S&P options tape properly — remove the matched call-and-put pairs that carry no direction, remove the deep in-the-money legs that are stock substitutes rather than bets — and what remains is strongly, unambiguously bullish: roughly a billion and a fraction of net directional premium, built almost entirely on put selling. On the face of it that is the single most constructive index print in weeks.

Then look at the calendar it was written on. Sort that same put selling by expiration and about eighty percent of it lives in November of this year, June of 2027, December of 2027 and December of 2028. The one near-dated expiration that governs this week is essentially flat — a rounding error of net buying. Somebody is willing to underwrite the S&P one to two percent lower for the next four months to two years. Nobody made a bet on Wednesday.

Both halves of that are information. The far-dated selling tells you that the institutions who would be forced sellers in a real dislocation are instead writing the dislocation to other people — that is a structural floor, and it is why the framework does not carry a crash thesis here. The flat near-dated book tells you the same institutions have no view on this week and are not paying to have one. Anyone who cites the bullish index flow as a call on Wednesday is reading the number and ignoring the date, which is the exact error this framework enshrined after being burned by it.

TAPE: S&P options, structure-stripped (matched legs and deep-ITM delta-one removed, 68% directional residue): calls +$278M net bought, puts -$868M net sold, true directional +$1,146M. By expiry, net put premium: Jun-2027 -$232M · Nov-20 -$218M · Dec-2027 -$140M · Dec-2028 -$134M · Aug-21 -$58M · Jul-31 +$1.3M. Nasdaq tracker true directional +$42M; S&P tracker -$22M (puts bought — the retail-facing venue is the one hedging).

What Actually Broke — The Build-Out Supply Chain, End to End

Name the casualties in order and the pattern names itself. The semiconductor tracker led it down -3.27%. Behind it: SanDisk -10.8%, Marvell -7.2%, Micron -7.0%, Western Digital -6.9%, Intel -7.9% after beating on every line 24 hours earlier, Taiwan Semi -2.9%, Broadcom -2.7%, AMD -3.3%, and the Korea tracker — functionally a memory-duopoly proxy — -6.3%. Alongside them the financing sleeve: Nebius -15.0%, CoreWeave -11.4%, Oracle -4.2%, Vistra -3.3%.

That is memory, storage, foundry, semi-cap, Korea, neocloud, power and funder credit. It is one trade with eight tickers. And the tell that it is positioning rather than fundamentals is what happened at the other end: the one megacap with no capital-spending burden at all closed +3.5%, and the software complex — the group that has spent nine months being told artificial intelligence would eat it — closed green with the cleanest institutional options signature on the board.

Intel deserves its own line because it is the cleanest illustration of the earnings regime this framework has tracked for six straight sessions. It beat on every line, spiked +12% after hours, faded to +4.6% by Thursday’s close, and then gave the whole thing back and more on Friday. A beat-and-raise round-tripping to below where it started is not a company problem. It is a regime: technology beats are being sold, receiver and defense beats are being bought, and the market is not paying for good news it thinks is expensive.

TAPE: Cohort closes 7/24 — SMH -3.27% · SNDK -10.79% · MRVL -7.21% · MU -6.99% · WDC -6.90% · INTC -7.89% · EWY -6.27% · NBIS -15.02% · CRWV -11.37% · ORCL -4.21% · TSM -2.93% · AMD -3.29% · AVGO -2.69% · VST -3.31%. Complement — AAPL +3.53% · NOW +7.44% · CRM +4.29% · IGV +1.01% · XLF +0.86% · XLC +0.87% · GOOGL +0.65%.

The Contradiction Worth More Than Anything Else on the Tape

While the semiconductor complex was being sold in public, it was being bought in the dark, in size, in daylight hours. The semiconductor tracker printed an institutional dark volume +717% above its own average — and the two largest non-auction dark prints on the entire market, across every ticker, were both in it: one and a half million shares near 560 at 15:48, and eight hundred and twenty-two thousand shares near 568 at 14:14. That is roughly $1.3B of genuine intraday demand accumulated during a -3.3% decline, and neither print is a closing-auction cross, which is the artifact that contaminates most single-day dark reads.

It is not isolated. SanDisk took a $285M block near 1425 at 15:21, on top of its auction print. Micron’s session shape was a reversal — it closed +1.9% off its low with a positive first leg — and its headline -$2.6B dark number is one closing-auction cross whose side tag is arbitrary by construction. Nvidia’s own pipeline verdict for the session reads, in the file, dip defended into weakness, with a positive final leg and a genuine $310M block at 14:59.

The correct inference is narrow and worth stating precisely: this is a reason not to chase the short, not a reason to buy the long. Someone with size is willing to own this cohort at these prices. That caps how far a momentum unwind runs. It does not tell you the unwind is over, and it does not make the names accumulate-rated — the framework demoted memory this cycle for exactly the two-session contradiction that produced it. Conflating “stop shorting” with “start buying” is how the last cycle got burned in the opposite direction.

TAPE: Largest non-auction dark prints of the day, all tickers: SMH 15:48 1.5M sh @ 559.91 = $839.87M; SMH 14:14 822.5K @ 567.74 = $466.97M. SNDK 15:21 200K @ 1425.39 = $285.08M AtAsk. NVDA 14:59 1.5M @ 206.52 = $309.78M AtAsk. QQQ 11:34 400K @ 689.70 = $275.88M AtAsk. XLF $140.72M + $120.68M two intraday buys. HYG 14:54 3.43M @ 79.23 = $271.46M.

Three Names Everyone Will Misread This Weekend

Apple’s “distribution” print was a financing leg

Apple closed +3.5% while its headline dark number was deeply negative and every intraday leg carried a bearish label — the setup for a hundred “institutions sold the rip” posts. Look at the same minute in the options file. At 15:02:53 Apple printed a September 330-strike call block of sixteen thousand contracts and a September 330-strike put block of sixteen thousand contracts, same second, same strike, opening. 16,000 contracts is exactly 1.6M shares of delta. The dark block at 15:02 was 1.6M shares. It is a conversion — stock against a synthetic, delta-neutral by construction, zero directional content. Reading the stock leg alone inverts the meaning of the day.

Korea’s collapse was a price event, not a distribution event

The Korea tracker’s -$712M dark number is 92% one closing-auction cross. The genuine intraday residue is under $60M. More telling: its options tape shows puts being sold nearly two-to-one against puts bought, with an 89% clean directional residue after structure is removed. Somebody wrote Korea downside into a -6.3% decline. That does not rescue the position — the invalidation broke and the watch is closed — but it changes what the decline means.

Intel’s dark tape said nothing; its options tape said accumulation

Intel’s dark number is 96% the closing cross. Strip it and the genuine dark flow is roughly flat. Meanwhile in options: calls bought and puts sold, both legs pointing the same way, for the second-largest positive structure-adjusted print in the single-name file — on a day the stock fell -7.9%. Somebody is accumulating Intel into the giveback of its own beat.

TAPE: AAPL 15:02:53 — Sep-18 330 calls 16,000 lots $27.76M + Sep-18 330 puts 16,000 lots $20.32M, same second, OPENING, prior OI 5,000; matching dark block 15:02 1.6M sh @ 332.78 = $532.45M. EWY structure-adjusted directional +$8.1M (puts sold $16.5M vs bought $9.0M, 89% residue). INTC structure-adjusted +$22.4M (calls bought $57.4M, puts sold $60.8M vs bought $30.1M, 67% residue).

Unusual Activity — Six Structures That Do Not Mean What Their Top Leg Says

1. The put ladder — $154M of index downside written in a single second

At 13:10:42 one participant sold S&P puts across four separate expirations simultaneously: August seventh at the 7300 strike, August twenty-first at 7250, September eighteenth at 7350 and November twentieth at 7300. Two of the four opened brand-new risk. All four strikes sit one to two percent below Friday’s close. This is the trade of the day and the engine of the entire bullish index number — a single desk underwriting a shallow decline out to November, at strikes that would only pay if the market broke below the bottom of its recent range.

2. The multi-year risk reversal — selling 2027 to buy 2028

At 11:49:15, roughly $187M of long-dated puts sold across June 2027 and December 2027, financing about $50M of far-out-of-the-money calls bought in June 2028 and December 2027. That is a professional expressing a multi-year upward drift by writing the downside to pay for the upside. It is the same conviction as the put ladder, at ten times the horizon.

3. The SanDisk straddle — $88M betting on movement, not direction

At 15:21:50, 2,000 August-nineteenth 1500-strike calls and 2,000 identical puts printed in the same second as a matched pair — a pure volatility structure, no delta-one content — with zero prior open interest on both legs, at implied volatilities near 141. On the day the stock fell -10.8%, somebody put on $88M of pure volatility. They are not betting on the direction of memory. They are betting the argument gets resolved violently by mid-August.

4. The Nvidia combo that is not a call buy

At 12:02:04 a twenty-thousand-lot September 200-strike Nvidia call block printed as a block trade — $40M, the kind of print that generates headlines. In the same second a twenty-thousand-lot September 200-strike put printed. Same strike, same size, same second. It is a combo. There is no direction in it. This is precisely the failure mode this framework rebuilt its options discipline around after mis-reading a similar package in July.

5. The Nov-20 four-legged package that reads as a $132M call buy

At 12:28:09 the S&P tape shows $132M of November 7000-strike calls bought — the largest single premium print in the file. In the same second: 7000 puts sold, 8000 calls bought and 8000 puts sold, all 2,000 lots. Four matched legs, one financing package, roughly $274M of gross premium and no directional content whatsoever. If you sort the day’s flow by premium and read the top line, this is what you get, and it is wrong.

6. The genuine bears — and they are all dated the seventh of August

Two prints on the tape are unambiguously, structurally bearish and both open new risk. At 15:58:26, two minutes before the close and after a -15% decline, 20,282 Nebius August-seventh 170-strike puts were bought as a block against prior open interest of about 1,500. At 14:20:18, 9,894 Bloom Energy August-seventh 192-strike puts — at the money — were bought against prior open interest of one. Add Iren’s August-seventh 34-strike puts. Three separate institutions bought at-the-money protection on the AI power-and-financing complex, all expiring the same day, all opening. That date is not decoration; it is the answer to “when.”

TAPE: 13:10:42 SPX puts SOLD — all outright legs, no matched call+put pairs and no deep in-the-money delta-one content in the package: Aug-07 7300 2,900 lots $13.01M OPENING · Aug-21 7250 4,000 $26.02M OPENING · Sep-18 7350 2,700 $36.10M · Nov-20 7300 4,000 $78.66M = $153.79M. 11:49:15 Jun-27 7925P $92.15M + 6800P $52.93M + Dec-27 5500P $16.63M SOLD vs Jun-28 9050C $41.44M + Dec-27 10000C $8.31M BOUGHT. SNDK 15:21:50 Aug-19 1500 straddle 2,000x2, $88.17M, OI 0. NVDA 12:02:04 Sep-18 200 combo 20,000x2, $58.0M. NBIS 15:58:26 Aug-07 170P 20,282 lots $32.65M OPENING. BE 14:20:18 Aug-07 192P 9,894 lots $30.72M, prior OI 1.

The Volatility Surface — Who Paid Up and Who Did Not

The single most useful discipline this framework runs is refusing to call anything cheap without pricing it in volatility terms, and Friday is a textbook case of why. Every large institutional protective print on the tape cleared at 0.94x to 1.06x of its own underlying’s at-the-money implied volatility — the S&P tracker’s 740 puts at 0.99x, the Nasdaq tracker’s 690 puts at 1.01x and its August 680 puts at 0.99x, Nebius’s 170 puts at 1.06x even against a base volatility above a hundred and sixty, and the software tracker’s at-the-money 88 calls at 1.00x. The professionals bought the body.

Every crowded lottery strike cleared at 2x to 3x its own at-the-money. The space-lottery 330 calls that traded 97,284 contracts at $0.12 priced at 3.10x. The credit crash wing at 2.36x. The semiconductor tracker’s 385 and 390 puts at 2.19x and 2.14x. The S&P’s 6000 put at 2.02x. Micron’s 500 and 520 puts at 1.81x and 1.73x. A $0.12 option is not a low-cost option. In volatility terms it was the dearest contract in the file. If you want the tail view, express it near the money and finance it by selling the wing the crowd is bidding — that is what the institutional prints on this tape actually did.

And one structural change matters more than any single strike. The S&P’s at-the-money term structure is no longer in punishing contango — it is kinked around three discrete events. There is a bump at Wednesday’s Fed, a second at Friday’s month-end expiration, and a third on the seventh of August that is larger than the Fed’s. Between them the curve troughs on the third and fourth of August. From the day after the Fed out to August expiration the total carry is under a vol point across three weeks — effectively flat. Two weeks ago that same curve was punishing. Owning convexity through this event window costs far less than it did, which changes the calculus from renting protection daily to owning a defined window. The market has told you where the window is: the seventh of August.

TAPE: Skew ratios (contract IV ÷ same-symbol ATM IV) — PROFESSIONAL: SPY Jul-31 740P 0.99x ($14.3M) · QQQ Jul-31 690P 1.01x ($35.8M) · QQQ Aug-21 680P 0.99x · NBIS Aug-07 170P 1.06x ($48.3M) · IGV Jul-31 88C 1.00x ($21.7M) · INTC Sep-18 150C 0.94x (BELOW ATM). CROWD: SPCX 330C 3.10x · HYG Sep 74P 2.36x · SMH 385P 2.19x · SPX Sep-04 6000P 2.02x · MU 500P 1.81x. SPX ATM term: 7/29 15.88 · 7/30 13.84 · 7/31 15.56 · 8/04 12.46 · 8/07 15.87 · 8/21 14.50. QQQ ATM 26.59 vs SPX 16.98; single-stock vol index 47.82 vs VIX 18.58 = 2.57x dispersion.

Why This Is Not About Iran

The Nasdaq fell on a day oil fell -2%. If the geopolitical channel were driving equities, that does not happen. The weekend brought a genuine de-escalation — strikes paused on both sides, futures up, crude down — and it matters enormously for oil, for the dollar and for the inflation path. It is second-order for the Nasdaq, which broke on Friday for reasons that had nothing to do with the Strait of Hormuz.

The more coherent ordering is that the pause is a consequence, not a cause. Consider what had accumulated by Thursday: a 10-year through 4.7%, crude up roughly a quarter in a month, national gasoline back above $4, the semiconductor complex — the sector responsible for the majority of the index’s earnings growth — down -16%, and October rate-hike probability north of three-in-four. That is a combination no administration wants compounding into an autumn. One independent technical read reached the same conclusion unprompted, attributing the pause directly to the bond breakout. The causality runs from the bond market to the geopolitics, not the other way.

And the bond damage is the part that does not reverse on a headline. Friday produced weekly closing breakouts on the two-year, the ten-year and the thirty-year at the same time — the thirty-year’s first weekly close outside a range it has held since October of 2023. A wick above a level is noise; a weekly close outside a three-year structure is a regime marker. The mechanism through which that hurts equities is not complicated: when a risk-free instrument pays more, profits sitting further into the future are worth less today, and the further out they sit the worse the arithmetic gets. That is precisely why the pre-profit, capital-intensive, growth-story end of the market broke first and hardest, and why the hyperscalers — who are now issuing debt to fund the build-out — got caught in the same net.

The one genuine relief in this week’s data is that the trend strength behind the move is decaying. The framework’s own trend-quality reading on the ten-year proxy fell sharply, and on crude it collapsed. The index trend readings improved in the same direction — the S&P tracker’s downtrend went from decisively reversed to effectively dead, and the same happened to junk credit, discretionary and communication services. A dead trend is not a bullish trend. It is an exhausted one. What is still decisively reversed is exclusively the Nasdaq complex — the Qs, the Nasdaq 100, the megacap basket and the leveraged Nasdaq product. That is the cleanest statement of where the damage lives.

TAPE: Trend-quality readings 0727 vs 0724 forward — S&P tracker -22.4 → 6.0 (reversed to dead) · junk credit -68 → -12 · discretionary -90 → -8 · comm services -68 → -2 · megacap basket -48.7 → -13.7. Rate proxy 106.6 → 80.3, crude 111 → 77.1, dollar 86 → 94 (strengthening). Still reversed: QQQ -7, Nasdaq 100 -7, megacaps -13.7. Dominant: real estate 112, inverse-Nasdaq 111, utilities 107, dollar 94.

The Timing Models — An Honest Fork

The two active projection variants agree on everything except the one thing that matters this week. Both had the mid-July peak resolving lower. Both have August resolving higher. They disagree on whether the low is already in.

• Variant B projected the trough in the 24th-to-25th window — which landed, arguably one session early, on Thursday’s close — and then projects an upward leg into a peak in the 28th-to-29th window. On this reading the market rallies into the Fed and fades after it. Timing: hit. Direction: pending, and Monday opens it.
• Variant A projected the peak in the 21st-to-23rd window — which also landed — and then projects the trough in the 29th-to-30th window, meaning the low comes at or just after the Fed, with the upward leg running into August. Timing: not yet. Direction: partially confirmed and unresolved.

The framework logged variant B as tracking last cycle and nothing in Friday’s tape breaks that — but a Monday gap that fades hands the read straight to variant A, and the framework declares the fork rather than picking the variant that flatters the rest of this report. Structurally both matched: decline from the mid-month peak, late-July trough, August recovery. Neither is a price target and neither is used as one — these tools give timing and shape; the magnitude comes from the expected-move bands, and only from there.

An independent technical read arrives at the same fork with different tools and lands on two candidate windows: the seventh of August and the twenty-eighth. Its test is the same one every other input on this desk is watching: bounce and fill the gap and you get the later date; keep dropping and you get the earlier one, with sharper corrections producing more meaningful bounces. That the option market builds its largest volatility hump on the seventh of August, that three separate institutions bought at-the-money protection expiring that day, and that an unrelated chart read names the same date, is the strongest cross-confirmation in this cycle.

Sentiment, Breadth and the Washout Case

The mood gauge printed 33.8 — deep in the fear band, second-lowest reading of the stretch, and up only fractionally on a session where the index was flat. The crowd did not un-scare. It is 18.8 points above the capitulation threshold that would trigger the framework’s contrarian bottom rule, so no formal signal fires — but the practical reading is that there is no euphoria anywhere on this gauge to fade, and disbelief has been this bull market’s fuel for three months. One caveat carried openly: no gauge was published on the seventeenth, so the five-day change cannot be computed this cycle and only the one-day move is cited.

The breadth picture is more interesting than the sentiment one. Technology’s bullish-percent index — the share of tech names on point-and-figure buy signals — closed at 29.58 after falling -4.55% on the day, with its momentum oscillator at 36.95. That index was above 80 in late June. On the chart it sits in the same band it occupied at the April 2022, October 2022 and March 2025 washouts, each of which preceded a durable bounce. Underneath the index, the equal-weight S&P showed more strength than the cap-weighted one on Friday — the signature of rotation rather than distribution.

Volatility is behaving like a market that has not panicked and does not expect to. Spot volatility closed at 18.58, below 20, having been repeatedly rejected at 20 by premium sellers; the three-month-to-spot ratio sits just above 1.10, mild contango with no inversion, and inversion is the stress line that has not been touched. Meanwhile the single-stock volatility index closed near 47.8 — near its high for the year, against an index volatility near 18.6. That ratio is the whole story of this market in one number: the index is calm and the constituents are not.

TAPE: Sentiment 33.8 FEAR, 1-day +2.8, 5-day UNAVAILABLE (no 7/17 gauge). Tech bullish-percent 29.58 (-4.55% on the day), oscillator 36.95. VIX 18.58; 3-month/spot ratio 1.104; constituent-vol index 47.82. S&P daily EMAs 5 7449 / 10 7471 / 20 7477 — price below all three. One vendor’s regime matrix reads its most bearish cell (below gamma flip, declining 10-day slope).

The Week — One Policy Binary and Four Capital-Spending Binaries

This is why this is a special edition. Monday brings durable goods and the Dallas Fed. Tuesday morning is Boeing, Coca-Cola, UPS and PayPal; Tuesday night is Bloom Energy, Seagate, KLA, Visa and Teradyne — which means the semi-cap read and the AI-power binary both start before the Fed. Wednesday morning is SoFi, Vertiv, ADP and Procter & Gamble; the Fed decides at two Eastern with consensus at no change and a second press conference from a chair the market is still learning to read; Wednesday night is Microsoft, Meta, Arm, Qualcomm, Lam Research and Robinhood. Thursday morning brings core inflation, the advance second-quarter growth print and claims; Thursday night is Apple and Amazon. Friday closes with employment costs, Chicago activity and the energy majors.

The framework has no edge on what those companies report. It has a well-tested edge on how this tape reacts. Six consecutive sessions of evidence say the same thing: technology beats get sold, receiver and defense beats get bought, and the specific line that decides it is capital spending. Alphabet beat and raised and lost -7.1% because the capex line went up and free cash flow went negative. Intel beat on every line and gave the whole thing back. Lockheed, RTX, Honeywell and General Dynamics beat and were bought with both hands. The trade on Wednesday and Thursday night is not the number. It is the guide, and inside the guide it is one line.

What the options market is positioned for, name by name, going in. Apple carries the most constructive structure-adjusted print of the four megacaps — and it is the only one with no build-out burden. Amazon carries a positive print and a rising accumulation ladder. Microsoft is the quietest of the four: a genuinely slow tape where the labels are reliable, and they read sold-dominant, with a supply-heavy book and its nearest support marked as supply. Meta is the outlier and the most instructive — its structure-adjusted flow is flat, but the composition is call selling, and it is pinned between a seller at 595 and a buyer at 583 into an implied weekly move north of eight percent. Somebody is writing upside in Meta into its own print.

TAPE: Structure-adjusted directional prints into the earnings block — AAPL +$18.1M (51% residue) · AMZN +$22.6M (80%) · MSFT +$4.9M (72%) · META -$1.9M (81%, calls sold $51.4M vs bought $42.4M). Implied weekly moves: MSFT ±7.5%, META ±8.6%, AMZN ±7.3%, AAPL ±4.5%, SMH ±7.6%, MU ±12.6%. Fed consensus 3.75% unchanged. Core inflation Thursday, forecast +0.1% monthly, 3.7% annual from 4.1%.

Tier Board — What Changed

• Memory and storage — DEMOTED from hold to reduce-into-strength. Two sessions of contradiction, with the pipeline’s own ladder-contrast and ladder-decay warnings firing on the same file. This is the demotion the framework was late to and it is taken on two sessions, not one.
• Korea — WATCH CLOSED, invalidated. The shelf broke by more than four percent in one session. Re-entry requires fresh two-session confirmation, not a bounce.
• Defense and build-out receivers — TIER 2 HOLD, second confirmation. RTX, General Dynamics and Lockheed all confirmed with rising ladders at new cumulative highs. Honeywell excluded this cycle on its giveback.
• Financials — PROMOTION CANDIDATE, watch with permission. Two sessions of the largest positive sector institutional flow plus a rising ladder plus a green tape. A third session assigns the tier; none is assigned yet.
• Software — avoid-pocket REMOVED, moved to neutral-watch. Retired not because one green day overrides a tag, but because the tag itself was built on a misread structure. That distinction is the honest one.
• AI build-out financing sleeve — NEW AVOID. Nebius, CoreWeave, Iren, Bloom, Applied Digital. This is where the genuine opening institutional protection sits and it is dated the seventh of August.
• Compute leaders — TIER 1 HOLD, trail unchanged. Nvidia’s dip-defended verdict and its 200 line remain the whole trade.
• Megacap funders — WATCH into the prints, absorb rather than chase. Unchanged. Boeing stays a stabilization watch with no pre-print position ahead of Tuesday morning. Metals stay blocked. Crypto stays a weak leg with no dip-buy.

Bottom Line

Friday was one factor being unwound while the rest of the market was bought, and the two halves netted to zero at the index — which is why the S&P finished unchanged on a day the semiconductor complex lost three percent. The damage is real, it is concentrated in the AI capital-spending supply chain, and it is not finished. But it is a factor event, not a market event, and the difference matters more this week than it has all month.

The mechanical setup into Monday genuinely favours the upside, and it favours it for reasons that have nothing to do with optimism. The Nasdaq walked itself underneath roughly a hundred and forty-two thousand contracts of one-sided protection expiring this Friday; those contracts are now in the money and the dealers short them have to buy back hedges on the way up. The S&P’s equivalent wall sits one dollar overhead. Dealer gamma is negative above spot, which amplifies an up-move by construction. Sentiment is in the fear band. Technology breadth is washed out at levels that have marked bottoms three times in four years. The crowd has openly disclosed itself as short-the-pop. And the weekend delivered a de-escalation headline with futures gapping higher on it.

The condition on all of it is one word: hold. Every pop for three weeks has failed to sustain — that is the pattern, and one more failure makes this the fourth. The specific confirmation is not a touch of the Nasdaq’s 690 level but a session close above it; that retires the wall rather than merely testing it. If Monday’s gap fades before midday, the mechanical case is spent and the map reverts to the downside levels below.

And the risk is dated. The market has told you when it thinks the reckoning is. The index volatility curve builds its largest hump on the seventh of August — larger than the Fed’s. Three separate institutions bought at-the-money protection on the AI financing complex expiring that exact day. An independent chart read names the same date as one of its two correction targets. And the far-dated index put selling that makes the flow look bullish is written in November, 2027 and 2028 — a structural floor, not a call on Wednesday. Trade this week’s mechanics with this week’s size, and own the convexity for the window the market actually paid for. The term structure is nearly flat from the day after the Fed out to August expiration, which is the cheapest that window has looked in a month in volatility terms — a change from the punishing carry that made the same trade a losing subscription two weeks ago.

What would change the view. A Nasdaq close back below its lower expected band, or an S&P tracker close below the bottom of its zone, kills the squeeze case outright. A junk-credit close below its zone floor re-arms the whole bearish stack. A ten-year proxy close above the top of its zone puts the rate channel back in charge and everything else becomes secondary. On the other side, a Nasdaq session close above 700 retires the wall completely and the framework would treat the July drawdown as complete.

Top Trades to Follow

MECHANICAL SQUEEZE QQQ close above 690 retires ~142K contracts of in-the-money one-sided put OI (19.8x / 18.5x / 8.5x at 690/695/700). Express near-the-money — the pro prints cleared 0.99-1.01x ATM IV. Target 700; invalidation a close below 674.98.

THE PUT LADDER 13:10:42 one desk SOLD $153.79M of SPX puts across Aug-07 7300, Aug-21 7250, Sep-18 7350, Nov-20 7300 — two OPENING. Those strikes are the institutional floor map, 1-2% below spot. Follow the strikes, not the premium.

AUG-07 AI-FINANCING PUTS NBIS Aug-07 170P 20,282 lots $32.65M OPENING (IV 175 vs ATM 165 = 1.06x, at par) + BE Aug-07 192P 9,894 lots $30.72M (prior OI 1) + IREN Aug-07 34P. Three institutions, one date. This is the dated risk.

SNDK VOL, NOT DIRECTION Aug-19 1500 straddle, 2,000 lots each leg, $88.17M, OI 0 both sides, IV ~141. Somebody expects memory’s argument settled violently by mid-August. Watch it as a clock, not a direction.

SEMI DARK ABSORPTION SMH 15:48 1.5M sh @ 559.91 = $839.87M + 14:14 822.5K @ 567.74 = $466.97M, both intraday, both non-auction, on a 717% above-average dark day. Reason not to chase the short. NOT a long signal.

META CALL SELLING Structure-adjusted -$1.9M on 81% residue, composition calls SOLD $51.4M vs bought $42.4M, pinned 583-595 into a ±8.6% implied weekly move. Somebody is writing upside into the print.

INTC ACCUMULATION Calls bought $57.4M + puts SOLD $60.8M vs bought $30.1M = +$22.4M structure-adjusted, 67% residue, on a -7.89% day. Plus Sep-18 150C at 0.94x ATM IV — below at-the-money vol, 62% out of the money.

AAPL CONVERSION — NOT DISTRIBUTION Sep-18 330 C+P 16,000 lots each, same second, OPENING, against a 1.6M sh dark block at 332.78 in the same minute. Exact delta match. Zero directional content — do not read the stock leg alone.

Key Levels — The Week

• S&P tracker. Overhead: 740 the in-the-money put wall and first hurdle, then 742, 744.65 the daily band, 748, 753.32 the weekly band, 755. Below: 737 the positive-gamma shelf that held all Friday, 735.33 the zone low, 733.21 the daily band, 727.49 the two-sigma tail.
• Nasdaq tracker. Overhead: 690 then 695 then 700 — the three retiring put strikes, and the confirmation is a close above the first. 693.48 the daily band, 706.80 the weekly. Below: 680 the next genuine shelf and August’s biggest one-sided strike, 677.98 the zone low, 674.98 the daily band, 665.73 the tail, 660 the September shelf.
• S&P index. 7395-7405 the positive-gamma shelf. Overhead 7468.93 daily, 7500 the largest gamma concentration, 7525.88 two-sigma, 7555 weekly. Below 7355.03 daily, 7350 the largest cumulative flow level that held Thursday’s low, 7298.08 two-sigma, 7268.53 the weekly floor untested since the range began.
• Small caps 288.18-294.16 daily, 283.92-298.42 weekly; the August 285 strike is genuinely one-sided.
• Credit — junk floor 79.18; a close below re-arms the bear stack. Rates — ten-year proxy zone top 47.35; a close above puts the rate channel back in charge. Dollar 101.63 zone top. Crude 86.07-94.87 daily. Long bonds sit under one percent above their quarterly floor at 82.50 — the tightest quarterly-floor proximity on the board.
• Nvidia 202.11-211.57 daily, 196.58-217.10 weekly; the 200 line is the market’s pivot as much as the stock’s. Nasdaq 100 28,500 is the level one desk called the critical threshold; Friday closed below it and the overnight bounce reclaimed it.

Sources

Expected Moves & Zones: daily expected moves 0727 · daily expected moves zones 0727 · daily range & trend 0727 · ZONE DOCUMENT 0727 · Zone Visual 0727 · weekly expected moves 0727 to 0731 · monthly expected moves July 2026 · quarterly expected moves July to September 2026 · JPM Collar levels Q3 2026 · FOM sentiment index 0724 · sentiment_index_tracker.md
Tradytics Dashboards & CSVs: options dashboard 0724.pdf (21 pages, read page-by-page: Market Net Flow · 0DTE Flow & GEX SPY/SPX/QQQ · Market DEX · Flow Map by Expiration · Flow Timeline · Dealers Diary · Top Flow · Call/Put Chains · Highest Call/Put Vol Change · High-Volume Cheapies · High-Volume LEAPs · Most OTM Strikes · Large OTM OI · Weekly Sector Inflow · Sector Flow · Sector Flow Premiums · Calls/Puts Market Dashboard) · darkpool dashboard 0724.pdf (10 pages: Stats · Live Darkpool Trades · Block Trades · Largest Darkpool Trades · Sector Darkpool Amount & Net · Darkpool Ticker Dashboard) · Live Options Flow - 0724.csv (33,801 prints, $13.33B gross — three-layer structure decomposition + implied-vol surface computed) · Darkpool Market Summary 0724.csv (3,288 rows, auction-cross stripped)
Timing: savino July 2026 projection version a 0722 update · version b 0722 update · inverse 0722 update (logged, not integrated) · Savino ZB_F bond forecast 0626 (stale, not integrated)
Commentary: mike silva - fom - Stock Market Report - 7_24_2026.pdf (49 pages; read: title, earnings calendar, INTC, MU, NVDA, SPY weekly-EM, macro calendar, market-conditions matrix, VIX/VIXEQ, VXV:VIX, USD/TNX/SPX, SPX daily EMAs, tech bullish-percent) · MAV 0726 · Cheddar Flow 0727 · ET Tradytics 0727 · Geeks of Finance 0726 · Mike Jones 0726 · Trading Apologist 0726 · VolSignals 0724 · Cem Karsan 0723 (tastylive)
Recon Pipeline: recon_data/2026-07-24/wl1/analysis_results/maverick_summary_2026-07-24_wl1.md (498-ticker census) · ticker_reports/ — 50 files read in full: SPY, QQQ, IWM, DIA, SMH, IGV, XLK, XLI, XLE, XLF, XLU, XLV, XLY, XLC, HYG, TLT, GLD, USO, MSFT, META, AAPL, AMZN, GOOGL, TSLA, NVDA, AMD, AVGO, MU, TSM, SNDK, WDC, MRVL, INTC, ORCL, CRM, NOW, BA, RTX, LMT, HON, GD, EWY, MSTR, IBIT, COIN, NBIS, VST, PLTR, CRWV, HOOD
Framework Context: comprehensive_analysis_0723.md · daily_report_0723.html · regime_snapshot.md (0723) · AN_FLOW_TRACKER_ROLLING_0723_v55.md · COMP_FILE_STRUCTURE.md
External / user-supplied events layer: US-Iran strike pause (weekend 7/25-7/26) · WSJ report of Nvidia financing talks for an Ohio OpenAI campus · Nvidia and SK Group HBM4 letters of intent · FX Evolution weekend recap · QQQ monthly seasonality table 2010-2026

Anti Narrative Platform — institutional flow analysis. Published 07/27/26 pre-open, on 07/24/26 closing data. This is analysis, not investment advice. Positions described are inferred from institutional flow data and may be wrong. Verification layer: the working analysis file for this cycle passed the inventory gate, the eight citation checks, the options-structure gate and the synthesis gate before publication.