Daily Report — 07/28/26 · “The Date Moved”
The chip complex printed its heaviest volume in roughly two years and the people with the longest clocks spent the day buying it. Two companies beat and raised and were sold anyway. The index bid that has answered every bearish input for six sessions finally broke. And the market quietly stopped paying for the crash date we have been building around since June, moved its money to August expiration, and extended its real protection out to December.
Scorecard — Grading Monday’s Report
“The Fourth Failed Pop” earns a B− on direction, an A− on its tags, and one outright retirement. Every cohort call it made paid: the memory and equipment reduce-tags had their largest day of the sequence, the AI-financing avoid-list produced an 11-point loss on a company that beat and raised, the financials hold got a fourth confirming session, and the instruction to stay flat on the biggest chip name was correct — it was the only green name in the entire complex. The texture call was right too: the report said thin gamma both ways, expect velocity, and the tape delivered a full overnight round trip inside the session and another one after the close.
What it got wrong was the shape of the forecast, not the content. The report gave Tuesday one distribution — coil, drift-up, flush — and the session printed two of those buckets at once. The index took the drift-up path. The factor took the flush. When our own convergence count reads zero, a single index distribution is the wrong instrument, and that is a structural fix, written up below.
And one standing assumption is retired today. Since June this desk has paid rent to August 7th as the market’s own crash window, on four consecutive readings of a bump in the S&P volatility curve at that date. The fifth reading killed it. August 7th now prices below the expiry two days in front of it, while August 21st repriced up by a point and a half and August 28th became the most expensive date in the month. In the same session one desk sold roughly a hundred and nineteen million dollars of Nasdaq puts expiring August 7th, in three strikes, in one second. Two independent instruments, one direction. The date moved, and our map moves with it.
Monday’s Top Trades, graded:
- Financials T2 — RIGHT, fourth session. Largest positive institutional net again, and the sector ETF was one of the day’s stronger performers. Caveat below: the names inside it changed.
- Memory reduce — RIGHT, emphatically. The storage complex lost between six and fourteen points. This tag has now paid four sessions running.
- Financing-sleeve avoid — RIGHT. The whole sleeve bled, and the fuel-cell name that beat and raised lost eleven points with a large put position opened into it.
- NVDA flat until the 200 line resolves — RIGHT. Unresolved, and the name outperformed every peer by holding green.
- Korea closed, don’t knife-catch — RIGHT. The country fund lost another six points on volume up three quarters.
- Quantum fade-rallies — RIGHT. Gave back the entire news pop.
- Crypto no-chase — RIGHT. Monday’s covering window closed in a single session.
- Software neutral-watch — PROCESS RIGHT, P&L WRONG. Two of the names ran four to five points while we declined to upgrade, because the confirmation we pre-registered still has not appeared. Third session of that. The discipline is intact; it cost money.
- Boeing no-position into its print — COST 4.8 POINTS. The multi-day pattern into it was distribution. No rule was broken. Note it and move on.
- The August 7th window — INVALIDATED. See above. This is the correction that outranks everything else in this report.
SCORECARD: Tuesday modal was Coil 45 / Drift-up 30 / Flush 25 — outcome DRIFT-UP at the index (SPX +0.21%, DIA +1.08%), FLUSH at the factor (SMH -3.45%, memory -6% to -14%). The 690-close squeeze trigger stayed unarmed. Structure-adjusted index options residue held at 59% of a $25.16B gross tape — readable, not structure-dominated.
The Read — Capitulation Volume, and Who Was Buying
The semiconductor complex had its heaviest volume day in roughly two years, and the professional side spent it accumulating. Storage lost fourteen points on its third consecutive double-digit decline. Memory lost nearly nine on volume that more than doubled. Equipment names lost six to eight. Every component of the chip index is now below its fifty-day average for the first time since April of last year, and the index is on pace for its worst month since 2008. This is what a bear market in a sector looks like while it is happening.
And every one of the worst-hit names printed net demand in its options book once the financing structures were stripped out. The single largest single-name trade of the day was a call bought at the ask in the storage name that fell fourteen points — dated to the middle of 2027, at an implied volatility roughly two-thirds of what the front month was charging. Put ladders were sold into memory, into two of the equipment names, and into the second-largest chip designer on their worst sessions of the cycle. Monday’s nine-figure Intel call position was added to on Tuesday, and the open interest at that strike is now roughly nine times what it was a session ago.
That is not what a demand break looks like. It is what a transfer looks like. Leveraged and retail-held inventory moved to somebody with a longer clock. Transfers take time and they do not bottom on the day the volume prints, which is why the reduce-tags stay in place. But it changes the instruction: the correct expression of a reduce-tag here is absence, not a short. Being right about the direction and wrong about the counterparty is how a purge takes money from both sides.
TAPE: Structure-adjusted single-name options residue — SNDK +$170.4M (86% directional residue) on a Jun-2027 1020C $124.5M at the ask, 117 vol vs a 183 front-month at-the-money · MU +$70.5M on $169.7M of puts sold · LRCX +$42.9M on a 3,600-lot deep-in-the-money put sold · AMAT +$38.3M · INTC +$47.9M, Nov-20 70C open interest 8,354 → 73,263.
The Beat Stopped Mattering — Two Proofs in One Session
SK Hynix grew operating profit five hundred and fifty-seven percent year over year, at a record operating margin, with memory pricing up thirty percent on the quarter and its next-generation high-bandwidth product beginning to ship — and the shares lost nine points, because revenue and profit both came in under consensus. Bloom Energy grew revenue a hundred and sixty-six percent, beat earnings by roughly ninety percent, and raised full-year guidance above where the street sat — and lost eleven points, with a twenty-five-thousand-lot put opened at the ask into it the same afternoon.
On the same morning, Coca-Cola beat and raised and rose five points. PayPal beat and rose four. Boeing beat and rose nearly five. That is the eighth consecutive session confirming the same reaction function, and Tuesday sharpened it: the discriminator is no longer the beat, and it is no longer even the guide. It is whether the guide implies more capital spending. A raise that means more capex is now read as a cost, not a signal.
Which makes one position in Tuesday’s tape the most informative thing in the file. Ninety minutes before the largest social platform reports, a sixteen-thousand-lot call struck well above the market and dated to January 2027 was sold at the bid, in a single block. Long-dated upside monetized on the eve of a binary the options market is pricing at roughly nine to ten percent on the week. Somebody with size does not expect this print to re-rate the name higher.
TAPE: META structure-adjusted -$51.7M on 75% directional residue — Jan-2027 750C ×16,000 $40.9M sold at the bid, 15:32 block, against open interest of 217,641 · front-week at-the-money volatility repriced 95.30 → 111.37 in one session · BE structure-adjusted -$33.8M, Aug-21 130P ×25,000 $31.9M bought at the ask, 173 vol against a 318 front month.
The Index Bid Broke
For six consecutive sessions this desk answered every bearish input the same way: the institutional tape kept absorbing at the index level. On Tuesday it stopped. The S&P tracker traded ten and a quarter billion dollars in the dark, up more than forty percent on the day, and nearly three-quarters of it printed on the bid — a net of roughly minus four and a half billion. The largest S&P mutual-fund-class tracker printed negative too. Only the third one leaned the other way, and by an order of magnitude less.
The reason this counts, rather than being waved off as a labelling artifact, is our own rule. We ignore at-bid and at-ask labels when the tape is fast, because in a fast tape they are mechanical noise. Tuesday’s index tape was slow — a quarter-point close — and in a slow tape our own reliability table says those labels are trustworthy. The symmetric obligation of that rule is that when the tape is slow, a four-and-a-half-billion-dollar bid-side net has to be counted. It is counted here.
Breadth agreed. Our five-hundred-name count went from fifty-four percent bullish on Monday back to forty-two percent on Tuesday — a complete round trip in one session, because the semiconductor complex is numerous and it all went the same way. That does not make the tape bearish. Four sectors still printed positive institutional nets and the index closed green. But it does mean the bull case now rests on rotation alone, and rotation is a relative argument, not an absolute one.
TAPE: SPY dark $10.29B (+43.5% d/d), $7.47B at bid vs $2.82B at ask, net -$4.65B on a slow tape · IVV -$726M, VOO +$589M · breadth 214 bull / 229 bear / 60 neutral of 515, from 278/158/66 · distribution ladders 96 vs 82 accumulation, 56 suppressed on tag-versus-net conflict.
The Date Moved — Our Own Forward Map, Corrected
This is the most actionable thing that happened Tuesday, and it was invisible on the chart. Since June the S&P volatility curve has carried a bump at August 7th — that expiry priced above both of its neighbours, four readings in a row. We treated it as the market telling us where it expected trouble, and we built hedge timing around it. On Tuesday the fifth reading came in and the bump was gone: August 7th now prints below the expiry two days ahead of it. Meanwhile August 21st — monthly expiration — repriced up by roughly a point and a half, and August 28th became the highest reading in the month.
The flow confirmed it independently and in size. At 10:59, in a single second, one desk sold Nasdaq puts across three August 7th strikes in a three-two-one size ladder: roughly a hundred and nineteen million dollars of premium collected, the largest leg opening fresh against existing open interest. Someone monetized the near-dated wall rather than rolling it. Two instruments, one direction: the professional hedge book gave up the near-dated crash trade.
It did not give up hedging. It extended. The largest single one-sided open-interest structure now on the board is a December Nasdaq put strike carrying nearly two hundred thousand contracts against fewer than fourteen thousand calls — a ratio of roughly fourteen to one, which is a genuine one-sided position rather than the call-put parity that marks a financing box. The August shelf is intact too, and it got more lopsided: the S&P tracker’s August monthly downside strikes now carry ratios of twenty-five to one and higher with essentially empty call sides.
Read together: near-dated protection was sold, the August expiration window was kept, and the real book was extended out to December. That is not bullish and it is not bearish. It is longer. And it means anyone still renting protection to the first week of August is paying for a date the market itself has stopped defending.
TAPE: SPX at-the-money by expiry — 7/31 20.28 · 8/05 15.29 · 8/07 15.17 (was 15.67) · 8/21 15.37 (was 13.95) · 8/28 15.73 · 9/18 14.76 · QQQ Aug-07 puts sold 10:59:04: 660P ×65,817 $51.9M + 690P ×21,939 $47.9M + 645P ×43,878 $19.1M · QQQ Dec-18 660: 194,395 puts / 13,852 calls = 14.03x one-sided · SPY Aug-21 720: 54,472 puts / 2,125 calls = 25.63x; Aug-21 730: 63,644 puts / zero calls.
Where the Money Went — Rotation Goes Vertical
The bid side of Tuesday was not subtle and it was not defensive-only. Healthcare led with a two-point-plus gain and did it on real institutional volume — the largest managed-care name traded well over a billion in the dark, up more than one and a half times its normal pace. Staples, communications, materials, discretionary and financials all closed green against technology and energy red. The Dow gained a point while the Nasdaq lost one, which is the widest one-day divergence of the rotation so far.
The trend regime underneath went from suggestive to unambiguous. Materials, healthcare, staples, financials, retail, the Dow, real estate and industrials all now register dominant confirmed uptrends on our range measure — the broadest dominant cluster of this entire cycle. Meanwhile the technology sector’s own trend reading turned negative for the first time, which in our framework means the trend line itself is now pointing the wrong way and is no longer usable as a reversion target. The Nasdaq’s own trend reading has been effectively dead for two sessions. A dead trend is exhausted, not bullish. A reversed one is worse.
Software held its bid without earning an upgrade, for a third session. Two of the big names ran four to five points and the sector fund closed green, but the confirmation we pre-registered — call buying at the sector-fund level, priced near at-the-money volatility rather than out in the wings — has now failed to appear three times. Tuesday there was no material sector-fund options flow at all. Meanwhile the analytics name lost six points and the database name closed flat with a sold-dominant institutional book on a slow tape. Rotation parking is not a re-rating. We keep waiting for the tell.
TAPE: Sector institutional nets — Financials +$1.76B (4th straight largest positive), Materials +$0.72B, Real Estate +$0.17B, Industrials +$0.14B vs Energy -$2.44B, Communications -$2.31B, Staples -$1.39B. Technology screens +$1.30B on raw labels but -$7.46B once the at-ask prints on names that fell six to nine points are corrected for tape speed · UNH $1.18B dark, +155% volume.
Financials Kept the Crown and Changed Hands
The sector took the largest positive institutional net for a fourth consecutive session, and its internal composition inverted. The money-centre and payments complex — the largest bank, the two biggest card networks, the premier investment bank, the second-largest bank, the biggest broker — all printed negative nets. What absorbed the sector’s inflow instead was insurance, exchanges and market-data, alternative asset managers, and regional banks. The regional bank fund traded on volume up more than two and a half times.
That is why the tier stays at two and does not get upgraded. When a sector’s leadership changes hands inside its own confirmation streak, the streak is telling you about the sector, not about the names. The destination is right; the vehicle is being swapped underneath us, and an upgrade would implicitly bless a name list that the tape has just abandoned.
TAPE: Financials constituent nets — positive: BRK/B +$821M, COF +$363M, MRSH +$334M, SPGI +$302M, ICE +$288M, MA +$281M, ALL +$261M, USB +$203M, C +$201M. Negative: JPM -$535M, AXP -$479M, V -$461M, GS -$342M, BAC -$175M, MS -$175M.
NVDA, and a Correction to Our Own Wall Map
The market’s biggest chip name was the only green thing in the complex, and its own institutional book printed two billion on the bid. Price up a quarter point on a slow tape with a heavily bid-side dark net is contested, not accumulated. Its options told the other story: net positive once financing structures were stripped, with December 2027 calls bought in size for a second consecutive session. Nothing here resolves the 200 line, and tomorrow’s expected range straddles it, so Wednesday settles it in either direction without needing an outsized move.
One correction is owed, and it is ours. Both the Friday and Monday reports described the 200 strike in August expiration as a call wall sitting over the stock. It is not. Checking both legs at that strike shows call open interest and put open interest identical to the contract — a one-to-one ratio, which is the signature of a box spread, an institutional cash-financing instrument with no directional content whatsoever. The 200 level still matters, but as the weekly expected-move floor and the two-hundred-day average, not as dealer geometry. We built that error the same way the industry builds it: by reading a one-sided out-of-the-money open-interest panel that shows you only one leg of a two-leg structure.
TAPE: NVDA dark $4.96B, net -$2.05B at bid on a slow tape · structure-adjusted options +$26.7M on 61% residue; Dec-2027 210C ×9,615 combined $35.7M · Aug-21 200 strike: 92,641 calls / 92,641 puts = 1.00x — box financing, parity exact, not a wall.
Korea — A Zero-Cost Bet on the Wreckage
The country fund lost six points on volume up three quarters and the Hynix receipt lost nine on a five-hundred-percent profit increase. The position stays closed. But the options book did something worth recording: it built a December call ratio, buying roughly twenty-four thousand calls near the money against selling roughly forty-eight thousand struck sixty percent higher, in a one-by-two, for essentially zero net premium — both legs struck at effectively the same implied volatility, so no volatility edge was taken either way. All of it opened fresh against open interest in the low hundreds.
What that structure says is precise, and it is not “buy.” Somebody is paying nothing to own a recovery to a specific level by December, and is willing to be short above it. That is a professional’s way of expressing a bounded recovery view in a name being liquidated by leverage, not a signal that the liquidation is over. Informational; not an entry. The emerging-market hedge showed up where leverage unwinds actually land: thirty thousand December puts bought at the ask in the broad emerging-market fund.
TAPE: SKHY structure-adjusted +$1.4M on 98% residue — Dec-18 140C ×24,146 bought $78.8M vs 210C ×48,292 sold $78.8M, legs at 108.75 and 109.04 vol, all opening · EEM structure-adjusted -$29.9M, Dec-18 65P ×30,000 $19.7M at the ask · EWY options flat — the three-session put-selling desk did not press.
The Credit Spine Got Worse
The stress is still in the financing sleeve, not the market, and it deepened. The database giant’s five-year default protection pushed above two hundred and ten basis points — reported as the widest since the financial crisis — with the rating agency now at one notch above junk after cutting it on negative free cash flow driven by AI infrastructure spend. Global margin debt is reported near one and a half trillion dollars, which is the mechanism by which a Korean leverage cascade becomes a Tuesday in New York. High-yield credit itself remains flat and its range has normalised, so the market-wide gate is still only leaning, not armed.
On the other side of the same coin, retail is positioned into exactly the cohort that is purging. Newly opened retail call positions are running near fifty-five percent concentrated in big technology on a twenty-day average, close to a record and above the level that marked the 2020 rebound peak. That is the population being transferred out of on days like Tuesday.
Overnight Addendum — The Missiles and the Round Trip
After the close, Central Command reported that Iranian forces launched multiple ballistic missiles at US positions in the Middle East and that all were intercepted. Crude, which had fallen more than three points during the session on reports that Egypt, Qatar and Pakistan were working to revive the original memorandum between Washington and Tehran, rallied roughly five percent after hours on the escalation. Equity futures spiked with the earnings tape, put in a high roughly seventy handles above the cash close, and had given all of it back and more by midnight. That is the fourth consecutive session in which the largest move happened outside regular hours and was fully reversed before the open — which is the practical cost of a market with no gamma stop between the bands.
Two things follow. First, oil is now a two-sided tail rather than a one-way disinflationary tailwind: the crowded long has unwound, the eighty-dollar floor was tested intraday, and the geopolitical premium is being re-added at exactly the moment the Fed would prefer it were not. Second, the pattern of overnight moves fully reversing by the open means the entry that matters this week is at the cash open, not at the futures print — and that is doubly true on a Fed day.
Unusual Trades of the Day
1. The $124.5M SanDisk Bet on 2027
At 15:56, with the stock down fourteen points on its third consecutive double-digit decline, a single buyer took twenty-five hundred June 2027 calls struck above the market for a hundred and twenty-four and a half million dollars, at the ask. The implied volatility paid was 117 — against a front-month at-the-money reading of 183. That is long-dated exposure bought at roughly two-thirds the volatility the panic was charging, which is the opposite of chasing a wing. Around it, the same book rolled its January 2027 puts down: selling the higher strikes, buying lower ones. Someone is not exiting this name. They are re-striking a position and extending the clock by eighteen months.
2. The One-Second, $119M August 7th Put Sale
At 10:59:04, three Nasdaq put strikes expiring August 7th printed in the same second in a three-two-one size ladder — roughly sixty-six thousand, forty-four thousand and twenty-two thousand contracts — all sold, two of them opening against existing open interest. Roughly a hundred and nineteen million dollars of premium collected against the exact date this desk has been treating as the market’s crash window since June. The volatility curve moved the same way in the same session. When flow and surface agree, the date is no longer being defended.
3. The Zero-Cost Korean Call Ratio
Two clips at 14:10 and 14:13 built a December one-by-two in the Hynix receipt: roughly twenty-four thousand calls bought near the money against roughly forty-eight thousand sold sixty percent higher, both legs at effectively identical implied volatility, for net zero premium, all opening against open interest in the low hundreds. It is a bounded recovery bet financed entirely by a capped short tail — and it was placed on the day the underlying lost nine points on a five-hundred-percent profit increase.
4. The Bloom Energy Put Into the Beat
At 13:16, twenty-five thousand August monthly puts struck well below the market were bought at the ask for thirty-one point nine million dollars, opening against open interest of fifteen hundred, at 173 volatility against a 318 front month — body pricing on a longer tenor, not a post-print lottery ticket. This was the afternoon after the company beat revenue by roughly a third, beat earnings by roughly ninety percent, and raised full-year guidance. The stock fell eleven points. The put is the professional side agreeing with the tape and disagreeing with the press release.
5. The Meta Call Sale Before the Print
At 15:32, ninety minutes before the largest social platform reported, a single block sold sixteen thousand January 2027 calls struck roughly a quarter above the market for forty point nine million dollars, at the bid, against open interest of two hundred and seventeen thousand. That is long-dated upside monetized on the eve of an event the market prices at roughly nine to ten percent on the week — and it left the name’s structure-adjusted options book the most clearly bearish of any mega-cap in the file.
6. The Intel Position Gets Doubled Down
Monday’s largest print of the day was a hundred and ninety-three million dollars of November calls struck at a level the stock has not seen in months, rolled out of a September position. Tuesday, on a session where the stock lost nearly six points, ninety-two hundred more of the same contract were bought at the ask. Open interest at that strike has gone from roughly eight thousand to seventy-three thousand in two sessions. Whatever this is, it is not a hedge and it is not a trade — it is somebody building a position while the tape hands them cheaper entries, and the volatility paid, in the mid-eighties, sits below where the front month is marked.
7. The New December Wall
The largest one-sided open-interest structure on the board is now a December Nasdaq put strike carrying close to two hundred thousand contracts against fewer than fourteen thousand calls. The ratio — roughly fourteen to one — is what separates a genuine one-sided hedge from a financing box, where call and put open interest sit at parity. Every large S&P strike on this board is a box. This one is not. Protection did not leave the market on Tuesday; it moved out five months.
Commentary Check — The Two-Year Volume Print
FX Evolution called Tuesday the biggest semiconductor volume day in almost two years and reached straight past the chip story to the funding story. His analogue is August 2024, and his transmission is the yen carry trade and leveraged product unwinding in Asia — the same mechanism we are reading through Korean receipts and emerging-market put buying. He also carried a study of chip-index behaviour after twenty-percent drawdowns going back to the nineties: only about twenty-nine percent of comparable cases made a new high within nine months, with further capitulation common along the way. His volume observation and our flow decomposition are the same finding approached from opposite ends — heaviest volume in two years is the transfer. Where we agree, and it matters: this is a bounce setup, not a bottom.
Mike Jones added the mechanical level. The Nasdaq fund hit its twenty-week average intraday and bounced more than a point off it, and the second open gap was filled. So the day’s decline ended on a real level rather than in free fall — which is consistent with a session that closed off the lows even as the sector bled.
Tim Knight, bearish and confirming on the index, handed us two dated catalysts we did not have. The space name has earnings after the close on August 4th and a lock-up expiration on August 11th at which a fifth of the float unlocks — a smaller unlock than it would have been had the stock held higher levels. He raises the converse-of-sell-the-news possibility that the stock rallies on the unlock. He also reads crude’s failed breakout as a bullish tell rather than a bearish one, which the overnight escalation immediately tested. Both of those dates land inside the window where the volatility surface just moved its premium.
The Mega-Caps Walk Into Their Prints
Google is the cohort leader on both tape and flow, and is the only one of the five trading above its own weekly ceiling. Microsoft closed up a point with an ask-side institutional book and weak-positive options into tonight. Apple closed up on genuine options demand, with a January 2027 straddle opened on top of it — the parking spot is being held with a volatility position layered over it, which is a hedge against the parking spot itself becoming a source of funds. Amazon was flat, distributed on a slow tape, and carries two large flexible straddles into Thursday: the market is buying its move, not its direction. Meta is the exception and the exception is bearish, for the reason above.
Every one of these earnings binaries repriced upward in a single session. The front-week volatility on all four mega-caps rose between fourteen and seventeen percent on Tuesday alone. That has a mechanical consequence: whatever the prints say, a very large amount of premium gets destroyed on Friday morning, and the names that hold their levels through it will be the ones the rotation buys next.
Timing — The Trough Window Opens Tomorrow
The projection variant we resolved to on Monday is unchanged and its trough window opens with the Fed. That path called a peak in the July 21st to 23rd window, which hit, a decline leg into July 29th or 30th, which is now five sessions old and confirmed, and an August up-leg after it. Tuesday matched the structure again, and it did so in the form this cycle keeps taking: the index refused to break while the factor did the declining. The competing variant needed a rally into the 28th or 29th and has now diverged for a second session.
What did change is the hedge calendar underneath the path. Both variants still agree on an August up-leg, but the market is no longer paying for the first week of August. It is paying for monthly expiration and the week after. If you were sizing convexity to the old window, resize it — the framework’s dated map should follow the money, not the memory.
Tier Moves
- Financials — TIER 2 HELD, no upgrade. Fourth confirming session, dominant confirmed sector trend, but the internal leadership rotated out of the money-centres and into insurance, exchanges, alternatives and regionals. Confirmation belongs to the sector, not the name list.
- Healthcare / defensive quality — NEW, STABILIZATION WATCH. Two clean sessions of three with a dominant confirmed trend and real institutional volume behind the managed-care and pharma names. Promote on a third. We do not upgrade cohorts on rotation days.
- NVDA — TIER 2 HOLD-REDUCED, unchanged. One stabilizing session after an invalidation is a watch, not a re-upgrade. Re-upgrade needs a close back above 200 after the Fed; further reduction below 190.
- Memory, storage and semi-cap equipment — REDUCE CONFIRMED, fourth session, and DO NOT SHORT INTO IT. The tape is capitulating while the options are accumulating, and those two disagree by design during a transfer. The correct expression of this tag is absence.
- Software — NEUTRAL-WATCH, not upgraded, third session. The pre-registered confirmation has failed to appear three times running, and Tuesday produced no material sector-fund flow at all.
- AI-financing sleeve — AVOID confirmed and widened. The fuel-cell name’s eleven-point loss on a beat-and-raise, with a large put opened into it, validates the tag without ambiguity.
- Industrial receivers and defence — TIER 2 HOLD, fourth confirmation. Held through a purge session without leading. That is what the tag is for.
- Crypto — DOWNGRADED back to weak leg, no dip buying. Monday’s covering window closed in one session. No tier until two consecutive sessions.
- Korea — CLOSED, unchanged. The December ratio is informational, not an entry.
- Metals — long still blocked by the strong-dollar condition, fourth session. The dollar sits at the top of its zone with a dominant confirmed uptrend, and both metals funds obeyed it on Tuesday.
- Oil — no position either way. The crowded long has unwound, the eighty floor was tested, and the overnight escalation makes this a two-sided tail rather than a directional trade.
- Space — fade-rallies becomes DATED. An August 4th print and an August 11th lock-up now sit inside the re-priced window, and the institutional book underneath Tuesday’s gain was three-quarters on the bid.
The Map — Wednesday and the Binary
The Fed decides at two, the new chairman speaks after, and the two largest capex disclosures of the season land the same evening. Market-implied odds put no move around three-quarters, leaving a hike tail in the mid-to-high twenties. We have no edge on the decision. We have an edge on the reaction function, and it is worth stating before the fact rather than after.
A hold with a balanced press conference resolves into the capex-light side, not into the index. The cohort that has absorbed six sessions of rotation is the cohort that gets bought again, and the front-week premium that gets destroyed on Friday funds it. The confirmation trigger is unchanged and still unmet: a Nasdaq-fund session close above 690, not a touch. A hold with a hawkish press conference is the worst configuration on this board, because it removes the relief without removing the August hedges. A hike prints the flush, and there is no dealer stop between the bands until the S&P’s weekly floor — and the August shelf is already armed to carry the follow-through.
One mechanical level is worth more than any round number this week. The systematic-trend cohort’s short-term trigger sits just above Tuesday’s close. Below it, the one-week projection is a seller in every tape state. Above it, the one-month projection flips to a large buyer on an up tape. That level, not a chart round number, is where the mechanical flow picture changes sign.
Friday is its own event for the chip complex. Roughly two hundred thousand puts across two strikes expire that afternoon in the chip fund, with essentially empty call sides — a pin magnet into the week’s close regardless of what the Fed does, and now from a lower base than a session ago.
MAP: SPY expected range Wed 734.74-746.98, outer 728.62-753.10 · QQQ 665.84-685.14, outer 656.19-694.79, with the 7/31 680/695/700 put strikes now in the money overhead · SPX 7371.14-7486.42; systematic short-term pivot 7455, medium 7204 · SMH 7/31 expiry: 101,569 puts at 520 and 97,705 at 500, both with zero call open interest — genuinely one-sided, not parity boxes.
Bottom Line
Tuesday was the semiconductor complex’s capitulation-volume session, and the people with the longest clocks spent it buying. That does not mean the low is in — the historical record after drawdowns this size says chop and retests, not V-bottoms — but it does mean fresh shorts into the hole are now trading against size. The tags stay: reduce and avoid mean absence, not shorts.
The bull case lost its foundation this session and it is worth being precise about which one. For six days the answer to every bearish input was that the index tape kept absorbing. It stopped, on a slow tape where our own rules say the labels are trustworthy, and breadth round-tripped with it. Four sectors still printed positive nets and the index closed green, so this is still rotation rather than de-grossing — but the bull case now rests on rotation alone, and rotation is relative.
And the map moved. The market stopped paying for the first week of August, kept the monthly expiration window, and extended its real protection out to December. We were paying rent to a date the market has now abandoned; that stops today. Own the post-event window, not the event, and own it where the professionals moved theirs.
Into Wednesday: no index position, a cohort split that survived its heaviest test, one broken bull input, and a hedge calendar that just moved a fortnight to the right. Let the binary spend the front-week premium. Deploy into post-event weakness on the capex-light side. And watch one number more closely than the Fed statement itself — whether the chip complex can hold its weekly floor into Friday’s expiry. If it can, the transfer had a buyer. If it cannot, the transfer had a bag holder, and August will be about finding out which.
Top Trades to Follow
SNDK LONG-DATED Jun-2027 1020C ×2,500, $124.5M at the ask, 117 vol vs 183 front-month — the largest single-name print of the day, bought on a fourteen-point decline, with the Jan-2027 put ladder rolled down around it. Grade this on whether the storage complex holds its weekly floor into Friday.
THE NEW AUGUST WINDOW SPX at-the-money 8/21 13.95→15.37 and 8/28 15.73, while 8/07 fell to 15.17 — stop renting the first week of August. Post-event convexity belongs at monthly expiration and the week after, where the market itself repriced.
META PRE-PRINT CALL SALE Jan-2027 750C ×16,000 sold at the bid, $40.9M, 15:32 block vs 217,641 open interest — structure-adjusted book the most bearish mega-cap in the file into tonight’s binary priced near nine to ten percent.
INTC POSITION BUILD Nov-20 70C +9,200 at the ask, open interest 8,354→73,263 in two sessions — two consecutive sessions of accumulation at the same strike through a six-point decline, paid in the mid-eighties vol, under the front month. Structure-adjusted residue 67%; single-leg, no matched pair or delta-one financing behind it.
DECEMBER IS THE REAL BOOK QQQ Dec-18 660P: 194,395 puts vs 13,852 calls = 14.03x — genuinely one-sided, unlike the S&P strikes on this board which sit at call-put parity and are financing boxes. Protection moved out five months, it did not leave.
FINANCING SLEEVE, STILL BE Aug-21 130P ×25,000 at the ask, $31.9M, 173 vol vs a 318 front month — bought the afternoon after a beat-and-raise. The sleeve's avoid-tag is the highest-paying tag in the book.
SMH FRIDAY PIN 101,569 puts at 520 + 97,705 at 500, zero calls at either strike expire Friday's close — a genuine one-sided expiry rather than a parity box, and a pin magnet for the chip complex from a lower base.
Sources: Live Options Flow 0728 (35,319 prints / $25.16B gross — full three-layer structure decomposition: matched legs $5.02B, stock-substitute $5.41B, directional residue 59%; plus a second multi-leg strip for same-second same-size spreads; per-symbol side-adjusted nets; per-strike open-interest parity on both legs; per-expiry at-the-money vol surface) · Darkpool Market Summary 0728 (3,185 rows, full aggregation and constituent-level sector nets) · forward 0729 expected moves, zones, range & trend, zone document and zone visual + weekly 0727-0731 + July monthly + Q3 quarterly + JPM collar · FOM sentiment index 0727 CARRIED — no 0728 file was dropped, sentiment inputs are one session stale and were not counted · NO options dashboard and NO darkpool dashboard exist for 0728; no claim in this report depends on a dashboard panel · Savino July timing variants A/B (0722 update) + ZB bond projection · Market Commentary 0728: FX Evolution, Mike Jones, Tim Knight (tastylive), Ilya Spivak, Geeks of Finance, Dividend Data · Silva 0727 + MAV 0727 + ET Tradytics 0728 carried · recon 2026-07-28 wl1: 515-ticker census with per-ticker detail · prior cycle: the 07/27 analysis file, regime snapshot 0727, rolling tracker v56, session log 07/28 · user-supplied Tuesday event layer (SK Hynix and Bloom quarterly detail, CENTCOM report, FAA launch-waiver proposal, NVIDIA-Commerce meeting, frontier-model policy, Moonshot and Alibaba chip reports, ADP weekly hiring, Visa layoffs, retail call-share, chip-index and fund-flow statistics, Oracle credit-default pricing and downgrade, margin debt, Apple market cap, Goldman systematic-flow table, KOSPI chart, overnight futures chart). Analysis file: the 07/28 verification layer (inventory gate PASS, citation gate 8/8 PASS, structure gate PASS).