← Back to All Reports
SPECIAL WEEKLY EDITION · BUILT ON THE FRIDAY 09/18 QUAD-WITCH CLOSE · THE INDEX WENT NOWHERE ON THE WEEK · THE YEAR-HIGH DARK-POOL NUMBERS ARE THE CLOSING PRINT AND THE INDEX REBALANCE · THE FRONT OF THE VOLATILITY CURVE IS NOW THE LOW-PRICED CONTRACT · PROTECTION IS DATED SEPTEMBER 25, OCTOBER 2 AND NOVEMBER

Daily Report — 09/18/26 · Weekly Edition · “The Record Was the Closing Bell”

The week that carried the first rate hike since 2023 ended within a tenth of a percent of where it began. Three down closes into the decision, a Wednesday low, then a two-session lift that flattened the volatility index and stopped on last week’s closing level. No index made a record. What did print records were the net figures on Friday’s dark-pool table — tens of billions in single names — and this edition traces them, print by print, to the quad-witch closing auction and the quarterly index changes: one stock being added to the S&P 500, with index funds forced to buy it, shows as the largest seller on the board. From there it reads the week from what can be trusted — price, the index wrappers, the dated option book and the volatility surface — grades a week of commentary, updates the timing model, and marks what the next two Fridays are hedged for.

The week in one line — a round trip, not a record

The S&P 500 (SPX) finished the week where it started it: 7,650.50 on Friday against 7,656.98 the Friday before, a change of less than a tenth of a percent. In between sat the first rate hike since 2023, three straight down closes into it, a Wednesday low at 7,551.81, and a two-session reversal that took the volatility index from 17.71 to 14.81. No index made a record. The S&P is about 1.9% under its August 13 record close; the Nasdaq-100 gained 0.94% on the week and is still 3.3% under its June high.

Underneath, the week was narrow. The Nasdaq rose while the Dow lost 1.69%, small caps (IWM) lost 1.66%, the equal-weight S&P (RSP) lost 1.20% and the financials (XLF) lost 2.43%. What did make records this week was something else entirely: the closing-print dollar figures on the dark-pool panel. That is the first thing this edition takes apart, because read naively it says institutions bought $35B of one stock and sold $14B of another. They did not.

The record on the panel was the closing bell

Every year-high net value on Friday’s dark-pool table is the quad-witch closing print and index-rebalance notional, not an institution taking a side. The per-ticker print files carry timestamps, so each name’s day can be split into what printed before 16:00 Eastern and what printed at or after it. For the thirty largest positive and thirty largest negative names on the table, between 77% and 100% of the day’s value printed at or after the bell, at the exact closing price. Fifty-four of the sixty names have a print file and every one reconciles to its own daily total within a tenth of a percent; six (ILMN, P, BRK/B, EBAY, TAP, IVV) have no file and are not verified here.

The names line up with published index events. SPCX, plus $34.5B on the table, is the Nasdaq-100 float-cap step-up that executed at Friday’s close after two lockup expiries — a passive purchase outside estimates put at $15.5B to $22B; a single block of $14.81B printed at 152.71 in the first quarter-second after the bell, and the stock closed down 1.36%. ILMN and P join the S&P 500 on Monday and show as large positives; TAP leaves it and shows as a negative; SNDK, PANW and DELL join the S&P 100.

The clearest case is BE. It is an S&P 500 addition — index funds were forced to buy it at the close — and the table shows it as the largest negative on the board, minus $14.16B, one hundred percent of it at the bell and tagged at the bid. Its trading before the bell netted to minus $12M on $39M. The stock fell 5.39% on the day, having run about a third higher into the inclusion: bought ahead of the add, sold into it. The tag followed the price, not the forced buyers. BE ZONE GAP: no zone coverage on the Monday sheet.

TAPE BE $8.39B single print at 265.63, 16:00:03 ET, tagged at the bid; the day’s remaining prints before the bell $39M · SPCX $34.28B of $35.48B at or after the bell · MSFT minus $9.70B on the table, plus $328M before the bell

The same pattern runs through the rest of the table. NVDA shows plus $17.89B; before the bell its prints netted slightly negative on a stock that rose 1.34%. MSFT shows minus $9.70B; before the bell it netted positive. SKHY shows minus $8.32B on a day it rose 2.46%. The sector bars on the same page — technology plus $41.9B, energy minus $8.8B, industrials minus $8.4B — are the same closing arithmetic added up by sector. None of it is rotation.

What that does to the accumulation ladders

This is not a quad-witch oddity. It is how the single-stock feed is built every day, and it changes what a multi-day accumulation ladder is worth. Across 652 names and the fifteen ordinary sessions before Friday, the median stock had 96.4% of its daily dark-pool value print at or after the bell. The reported daily net carried the closing print’s side tag 98.1% of the time and agreed with the before-the-bell net barely more than half the time. Strip the closing print and 39% of the sixteen-session ladders change sign: INTC reads plus $14.64B as reported and slightly negative without it; MSFT reads minus $13.49B reported and slightly positive without it.

The honest strength of the finding matters, so here it is measured. The closing tag is not random: it agrees with the stock’s same-day direction about 60% of the time, with an average same-day return of plus 0.36% after a positive tag against minus 0.65% after a negative one. It is an echo of the day’s price. It carries nothing for the next day (51%). And the before-the-bell net, which would be the tempting replacement, carries nothing either — 48% same-day, 50% next-day — so this edition does not swap one number for the other. The sample is fifteen sessions in one regime and tests one-day horizons only; it does not settle whether a multi-day slope persists.

The practical consequence: price on volume stays the read for single stocks, and the ladder tags are quoted but not leaned on. The index wrappers are different — SPY had 18% of its value at the bell, QQQ 35%, IWM 37% — so their thick-tape readings keep their meaning. Friday’s wrapper tape was mixed and labels fought price on two of three: SPY net negative on a flat day, QQQ net negative on an up day, IWM net positive on a down day. No input there.

The regime under it — a hike, a bond that lost its line again, a dollar over 100

The Fed is tightening, the ten-year is back at 5.00%, and the long bond gave back in one session the line it reclaimed on Thursday. Treasuries (TLT) closed 81.25, under the 81.35 level that the prior edition built its headline on, and under their quarterly band floor at 82.50 for a second week. With real yields above breakevens, a ten-year at five percent supports a market multiple near 11x; the longest-duration equities feel that first, which is where NFLX, CRM and IBM sat this week.

The dollar index holds above 100 with a strong range reading, which keeps the strong-dollar block on new bullish metals exposure in force — a rule about which structures are permitted, not a forecast that gold falls. Gold (GLD) closed 401.17, inside every band. High-yield credit (HYG) at 78.53 sits only 0.18 above the line that would signal a credit crack; it is clear, and thin. Crude closed the week at 100.30 after a 105.83 high close on Tuesday.

The event card from the hike reads unclear and stays live through Monday, which means no index direction is licensed yet. The standing registered view — an S&P close under its monthly band floor at 7,428.73 by October 16 — is open and ungraded; nothing in this edition registers a second one.

The four timeframes — where price sits and what it caps

The S&P starts the week dead centre on every clock, which is a statement about range, not direction. Against Monday’s daily band 7,607.22 to 7,693.78, the new weekly band 7,551.84 to 7,749.16, the monthly band 7,428.73 to 7,943.55 and the quarterly band 6,929.27 to 8,069.45, the close of 7,650.50 is within three-tenths of a standard deviation of the middle of all four. The weekly lower band sits almost exactly on Wednesday’s closing low, so a second test of 7,552 is the week’s whole downside in one number, and with eight sessions left in the month the monthly floor at 7,428.73 is a full-sized monthly move away.

The zone sheet for Monday repaired sharply: the S&P range reading went from negative to 57 and the Nasdaq proxy’s (QQQ) from 9 to 102 in two sessions, with the S&P zone at 7,528.95 / 7,633.98 / 7,739.01. Trend validity is back on the index. It is not back in breadth. Small caps (IWM, 284.10) sit below the middle of their daily band 281.87 to 286.33, their weekly band 279.12 to 289.08, their monthly band 281.09 to 306.77 and their quarterly band 272.21 to 328.69, with a zone of 279.66 / 287.29 / 294.91 and a trend line above price. That is continuation, driven by the ten-year; the reversion trigger is a reclaim of the zone middle at 287.29, and until it prints nothing is sold against IWM at a band low.

The Zone Visual for Monday draws the same split in colour. The week’s leaders sit pressed against the top of their zones on rising trend lines — QQQ has 0.29% of room to its zone high against 2.54% to its low, the technology fund (XLK) 0.31%, ARKK 0.63% — while the laggards sit at the bottom of theirs on falling trend lines: XLF 1.22% above its zone low with 4.33% of room above, the Dow fund (DIA) 1.23% above its low, regional banks (KRE) 1.18%. The dollar index is at the top of its zone too, 0.38% from the high. For Monday that is an asymmetry, not a forecast: the leaders have almost no zone room left and the laggards have almost nothing but. One caution on the artifact itself: the Zone Visual’s printed room percentages were drawn off an earlier mark than the numeric zone document’s closing basis (it prints 0.59% of upside for QQQ where the document, which reconciles to the close, gives 0.29%), so the figures here are the document’s and only the trend colours and placement are taken from the Visual.

QQQ at 721.45 opens inside its daily band 715.85 to 727.05 and weekly band 709.26 to 733.64, just under the 725 strike where the market’s own October 2 call supply sits. Four names closed outside a monthly band and each carries a consequence. META (665.75, zone 602.38 / 659.48 / 716.58) is above its monthly two-standard-deviation upper at 663.82 after a down day: a reversion setup, triggered by a Monday close back under 663.82. AMD (559.82, zone 460.19 / 517.04 / 573.88) is above its monthly upper at 524.05: continuation, the input being the wafer price increase, graded on holding the weekly lower at 528.61. NFLX (71.79, NFLX ZONE GAP) is below its monthly lower at 75.30: continuation, the input being the rate de-rating of long-duration names, with reversion only on a reclaim of the weekly upper at 74.33. AAPL (336.13, zone 314.58 / 329.26 / 343.93) is a hair over its monthly upper at 335.12 on a down day: no basis either way.

Winners and losers — read from price, because the options tape netted to nothing

Across the fifty-six names with the heaviest options traffic, the week’s structure-stripped net premium was under a tenth of gross for thirty-nine and never above a third for any. Six end-of-day files run through the same decomposition give file-wide nets of a few tens of millions a day on about $5B of directional-eligible premium. The one outlier was Thursday, minus $313M, and it was put buying into the rally day — hedging, which in a hedged market is what a bounce looks like, not a tell against it. So the week’s leadership is read from price, with the setup tags beside it.

Crypto beta led, on a two-session sample

Bitcoin cleared 80,000 on Friday and the proxies went with it: MSTR plus 17.5% on the week to 153.92 (MSTR ZONE GAP; weekly band 140.97 to 166.87, and now near the top of its monthly band at 156.28), COIN plus 10.8%, HOOD plus 6.4%, the bitcoin fund (IBIT) plus 5.1%. Holders wrote calls into it — the monetization read is harvest on MSTR and IBIT, which caps upside into the next expiry and is never a short. Implied volatility rose with price on MSTR, COIN and IBIT. The measured history for MSTR after a day of plus 10% or more is a coin-flip next session and a median worst dip near 12% within ten sessions, with the ten-day return still positive. That argues for owning it through defined-risk call structures and harvesting strength, which is what the tape’s holders did, and against selling any call against it.

TAPE MSTR Sep-18 150 calls 103,588 lots and Sep-25 150 calls 41,176 lots on the chain panel; call volume about 180K, the top of the call-volume-change panel · IBIT Sep-25 46.5 and 48 calls 57,607 lots combined

Memory and AI hardware bounced hardest off the low

SNDK gained 17.9% from Wednesday’s close and 9.7% on the week (SNDK ZONE GAP), MU 9.6% off the low to 1,015.80 (zone 908.72 / 978.27 / 1,047.81; weekly band 958.17 to 1,073.43), AMD 8.5% on the week, INTC 5.5% to 108.60 (zone 88.19 / 101.85 / 115.50). NVDA was the quiet one: plus 1.8% to 222.27, inside every band (zone 207.98 / 219.28 / 230.58; weekly 214.48 to 230.06), with its fixed-expiry implied volatility down eight percent on the week. Two sessions off a low inside a flat week is a stabilization watch, not an upgrade. MU reports after the close on Wednesday September 30, and its October 2 options already price the event at 65.7 implied against 58.3 in November.

TAPE MU top of the net-premium panel at plus $66.6M; Sep-18 1000 calls 140,704 lots · NVDA plus $37.2M; Sep-18 220 calls 420,777 lots · SNDK plus $34.3M, Sep-18 1740 calls at 30.00 — all front-expiry, raw panel figures before structure is stripped

Cybersecurity led early and faded into the reversal

CRWD finished the week up 15.0% and PANW up 10.0% (both ZONE GAP), but all of it came before Wednesday: from the low CRWD lost 1.5% and PANW 3.2%. Software as a group (IGV) did the same, plus 2.8% on the week and slightly down off the low. The reversal’s money came out of what had already worked and went into what had been hit hardest.

The banks were the week’s clear losers

GS lost 8.5%, BAC 7.9% (zone 56.56 / 60.11 / 63.65, closing 57.73), MS 5.5%, C 5.1%, WFC 4.6%, and JPM 1.8% to 349.67 (zone 345.75 / 352.62 / 359.49). XLF did not bounce at all from Wednesday’s low while SPY gained a percent, and sits near its monthly band floor at 55.65 (zone 55.18 / 56.73 / 58.28). A hike that flattens the curve hurts the lenders, and a trading-revenue warning from BAC added to it. This is a bearish trend read on price across five sessions; the line that ends it is an XLF reclaim of its zone middle at 56.73. Bank implied volatility is the one corner of the surface that rose on the week, with JPM’s October expiry up eight percent into the October 13 reports.

TAPE Financial the only meaningfully negative sector on the average-net-premium panel at minus $0.87M per name · JPM Oct-16 implied 26.1 against 21.7 in the front week

Power and exploration fell; the refiners did not

NRG lost 8.6% and VST 5.2%; CRWV lost 8.6%. In energy the split was clean: refiners MPC plus 7.3% and PSX plus 5.3%, explorers COP minus 4.0% and DVN minus 3.2%, as crude came off its Tuesday high. The refiners carry a policy tail — one commentator is warning of a diesel export ban on a single unnamed source — and nothing in any file confirms it.

The small-reactor names were Friday’s hardest-hit corner: SMR minus 8.5%, NNE minus 7.3%, OKLO minus 4.2%. That is worth stating because the speculative-names monitor raised a “strength being sold” alarm on the session, and it should not be read that way. Its basket averaged plus 1.2% only because five crypto names rose eight to sixteen percent; the median name fell 1.4% and fifteen of twenty-seven were down. It was a split basket — crypto up, reactors, quantum and air-taxi names down — not strength meeting supply.

The mega-cap roster, by level

MSFT closed 493.78, down 0.4% on the week and mid-zone (482.46 / 495.60 / 508.74; weekly band 480.64 to 506.92). GOOGL 349.54 (zone 330.42 / 342.15 / 353.87) and GOOG 344.41 (zone 328.07 / 338.65 / 349.23) each gained about three percent; GOOG has no weekly, monthly or quarterly board row and is carried on its daily zone only. AMZN 253.71 lost 1.2% (zone 243.03 / 252.94 / 262.85). TSLA 364.27 was flat, sitting on its zone middle (350.78 / 364.87 / 378.95). AVGO 357.61 lost 1.2% (zone 332.87 / 353.53 / 374.18; weekly band 344.03 to 371.19). SPCX closed 152.71, up one percent on the week through its own index event (SPCX ZONE GAP; weekly band 144.24 to 161.18, monthly 126.50 to 160.88).

The setup scan agrees with the price map rather than fighting it: seventy names tagged as sponsored, three as abandoned (a description, not a call), and five carrying the one flow-visible bearish setup — a vertical run with call chasing and heavy call selling into it — on TEM, CLS, YPF, DINO and SEI. None is an index heavyweight.

The index options tape — still structure, still no direction

On the quad-witch the S&P options file was financing and expiry plumbing with a directional remainder too small to read. Of $1.48B of S&P premium, the structure-free directional residue was plus $31.7M, about two percent: structure-dominated, not a direction. The large December open-interest rows on the dashboard — 8,000 calls, 7,000 puts — are box spreads, with call and put open interest at the 7,000 strike within six percent of each other; that is a loan, not a wall. QQQ was 59% structural. The one genuinely one-sided strike on the board, SPY 760 with puts more than double calls, expired Friday.

Flow map, flow timeline, dealer book — what the three say together

The reversal was the front hedge netting out at expiry, and the protection that matters was re-dated rather than removed. The flow timeline’s September 18 line ran from about plus 39 to plus 131 in the last two sessions — a vertical move on a front-expiry line the day before and the day of expiry is the existing hedge coming off, not fresh buying. Over the same two sessions the October 2 line fell from about minus 28 to the minus sixties, and the September 25 line made a new low near minus 75.

The dealer book confirms the mechanics. Friday’s expiry carried roughly 27B of call delta and 8.7B of put delta; every row after it is under 2B. The net dealer delta reading for the day was about plus 2.1B, the largest since late August. The S&P closed inside a cluster of negative zero-day gamma between 7,635 and 7,655 with its largest positive strike below at 7,620, which fits a late lift rather than a pin; the strike labels on that panel are inferred from unlabelled rows and are approximate.

For Monday: the mechanical support that produced the two-day lift is gone, the residual dealer book is small, and front-week volatility is priced for very little. That combination usually means a quiet open that can move further than the front prices imply in either direction. For the week: the dated protection sits on Friday September 25 — which is also a crypto expiry cluster — and on October 2, the payrolls session, with the larger volatility structures dated to November, past the midterms. The flow map’s September 25 row is net long puts. The market’s own bracket for the turn of the quarter is roughly QQQ 690 to 705 below and 725 above.

Volatility — the front is now the low-priced contract

After the expiry the front of the curve is the low-volatility contract almost everywhere, which retires the sell-the-front calendar that fit last week’s tape. S&P at-the-money implied volatility is 9.3 for September 25, 12.0 for October 2 and 13.1 for November: a front-to-back ratio near 0.71x. Of ninety-five names measured front week against the next liquid back month, four had a front priced at 1.10x the back or more, and only one of those is liquid — BABA at 50.2 against 43.7, 1.15x, carrying weekend trade-talk premium. A week ago fronts ran 1.10x to 1.26x over the back across the board.

The wings tell the other half. The S&P five-percent put wing prices 2.12x its at-the-money in the front week against 1.30x in November; QQQ’s October 2 put wing is 1.58x against 1.18x in the October 16 expiry. So the body is low-priced in the front, the wing is where a buyer pays up, and the structures at the end of this edition are built that way round. One commentator’s view that midterm volatility is underpriced is a judgment, not something measured here; what the surface does show is where volatility is priced — November and December, at 13 to 14 against a nine handle this week.

Sentiment — the relief snap did not extend

The sentiment gauge printed 43.2, neutral, up less than a point on the day and 2.7 on the week. It went from 40.5 the prior Friday to a 26.3 low on the hike and back to the low forties — the same round trip as price. It is 1.8 points under the 45 line that would cancel the fear-band reading, far from the capitulation threshold below 15, and far from the greed extreme above 80. A crushed volatility index beside a mid-neutral mood gauge is neither euphoria nor capitulation, so sentiment adds nothing to either side this week. Retail surveys are more bearish than the gauge — one survey has bears at 53% — while margin debt is at what one commentator calls a record. Both can be true of a market that went nowhere.

The commentary week — who was right

By Wednesday night most of the shelf had called a breakdown; by Friday most had called a bear trap. The index closed the week flat, which means both camps got a piece and the narrow-breadth bears got more than the headlines suggest. Calls are graded on the closes, and the load-bearing ones were checked against the transcripts.

The volatility desks had the best week

Volsignals laid out the sequence in advance — sell first on the decision, then a mechanical slingshot into expiry — said 7,550 would hold with an automatic bid above it, and gave a Friday settlement range of the 7,630s to 7,675. All three hit. Beside them: 7,850 by Friday was floated as possible and missed by two hundred points, the Friday straddles did not pay, and the desk named its own miss on the size of the volatility spike. SpotGamma’s call that a 17 to 18 volatility index would come in toward fair value, and that technology would beat the S&P while what he counts as a record number of index members moved against it, both hit.

Karsan owned a miss and changed his map

He had leaned against a hike and said so plainly on Friday: they were wrong. His turn window landed on the right dates for the wrong reason, his warning that a flatter curve hurts banks and small caps was exactly right, and he withdrew his second window. His new view — dips bought into the midterms, a mark-up into quarter-end, weakness after — is pending. The dated flow agrees with him on where the volatility is and does not yet confirm a push through QQQ 725 before October 2.

The Maverick called the breakout in crude, not the pullback

On Monday he had been calling for crude back at 100 and leaned toward an explosive break of weekly resistance; it closed 105.83 the next day. That call hit. The pullback did not come from him: on Wednesday he described crude above 100 as a flag before another leg higher, and on Thursday he put the overnight drop down to the contract roll. Crude closed the week at 100.30. His standing bearish lean on the index missed through Friday, his warning on software missed, and his diesel-export-ban alert rests on one unnamed source with nothing on disk to grade. He is context to weigh, never a signal, and this week the weighing says: right on the barrel’s first move, early or wrong on the rest.

James leads on crypto and was wrong about records

The standing bull’s calls that the hike would not hurt bitcoin and that it would retake 80,000 both hit, and his read that crude looked toppy is right so far. His line that the week brought new all-time highs is not supported by any index. He leads the week on the asset he follows most closely; the broad-market claim does not survive the tape.

How a short book handled the squeeze

Tim Knight ran twenty-eight shorts at 170% commitment after the hike and added into Thursday’s rally to about 200%. His Thursday test — a red Friday would confirm a rejection — failed on the S&P and the Nasdaq and held only in small caps. On Friday he covered his largest position, the Dow fund (DIA), two hours before recording, trimmed to about 125%, and said of the cover, “I’m out, so if we crash on Monday, shame on me.” The irony is that his book was better than his mood: the Dow, small caps and financials were the week’s losers and those shorts worked. His calls for gold lower and bitcoin to 67,000 both missed. The lesson on the page is about sizing into a mechanical rally, not about being wrong on breadth.

Silva, and a correction

Mike Silva labelled conditions bear-strong as of Wednesday’s close — a falling ten-day average on the Nasdaq Composite with price under the gamma flip — while saying the week was near its low, which it was. By Friday the S&P was back over the flip; no later Silva file is on the shelf, so his current label is unknown. One correction to this edition’s own working: gold did fall on the decision (GLD minus 0.6% on Wednesday, plus 1.7% Thursday), as several commentators said; a futures settlement timed before the announcement made it look otherwise.

Timing — the clock is two for two, the sign is one for two

The updated September projection landed its second turn date, and its author reversed his orientation call for the second time this month. He had flagged about September 15 expecting a high; it was a low, printed on the 16th, one session from the drawn vertex, and he now watches whether that low holds through month-end. Scored by inflection: the turn of September 9 to 10 was a low, which the inverse chart had and the standard chart did not; the turn of September 15 to 16 was a low, which the standard chart had and the inverse did not.

Timing charts give timing and shape, never a price target. The levels come from the boards: if the late-month turn is a low, the S&P weekly lower at 7,551.84 and the zone low at 7,528.95 are in play; if it is a high, the weekly upper at 7,749.16 and the zone high at 7,739.01. The practical use is the date: Monday the 28th is drawn as the month’s most violent session on both charts, it follows a crypto expiry cluster on the 25th, and it leads into quarter-end on the 30th and payrolls on October 2. There is no bond timing chart on the shelf.

Unusual trades — the structures worth naming

HYG February 2027 80 calls — 71,000 lots opened

Two prints, 66,000 and 5,000 lots, lifted toward the ask at 0.12 against open interest of about 60,000, with the credit fund at 78.5. About $852K of premium for upside in high-yield credit five months out. It is priced at 3.9 implied against roughly 5 at the money in the back months, so this is a near-the-money body bought below at-the-money volatility, not a rich wing. Structure checked: single-leg, no matched leg in either second, and entirely time value, so it is not a box or a stock substitute. With HYG only 0.18 above its credit line, someone is paying a small premium for credit holding up into the new year.

TAPE HYG Feb-19-2027 80C 66,000 at 11:02 and 5,000 at 11:26, toward the ask at 0.12, implied 3.87 · Jan-15-2027 80C 10,000 at the bid at 0.05 the other way

CMCSA — a three-leg roll into at-the-money 23 calls, not a fresh two-expiry purchase

Three prints in the same second: 10,000 January 2029 23 calls at 4.48 and 5,000 April 2027 23 calls at 2.08, both toward the ask with open interest of one and zero, and 8,000 January 2028 27.5 calls at 1.85, mid-market and unsided, against open interest of 8,245. A leg sized almost exactly to its open interest is most likely the position being rolled from, so the structure reads as a holder moving out of out-of-the-money January 2028 calls into at-the-money calls both nearer and farther — roughly $5.5M bought against about $1.5M coming out, if that third leg was the sale. It is a matched multi-leg package, so no single leg is read as direction on its own; all of it is time value, none of it delta-one. CMCSA ZONE GAP on the numeric sheet (its row falls in a section the parser refused); the Zone Visual places CMCSA at the bottom of its zone, 1.35% from the zone low with 16.19% of room to the zone high on the Visual’s own mark — so the holder re-struck closer to the money at the low end of the range, not into a breakout.

TAPE CMCSA 11:17:47 · Jan-2029 23C 10,000 at 4.48 ($4.48M) and Apr-16-2027 23C 5,000 at 2.08 ($1.04M), toward the ask, opening · Jan-2028 27.5C 8,000 at 1.85 ($1.48M), mid-market, open interest 8,245

TLT September 2027 81 calls — a one-year duration call, opened the day the bond lost its line

3,480 lots at the ask at 3.85, open interest zero, $1.34M, at the money with TLT at 81.22 and 12.0 implied. Structure checked: a single print in its second with no matched leg, 94% time value, so not a delta-one stock substitute. Back-month rate volatility was the sticky part of the surface this week, so this buyer paid a firm price for a year of room rather than trading the weekly line. It is the opposite time horizon to the week’s argument about 81.35.

TAPE TLT Sep-17-2027 81C 3,480 at the ask at 3.85 at 11:34, opening, implied 12.01

FCX — a roll up and out that took money off, not new upside buying

The 5,128 January 2028 85 calls bought toward the ask for $6.49M sit on the net-premium panel’s positive side, and read alone they look like fresh long-dated commodity upside. They are one leg. In the same second 2,246 January 2027 38 calls were sold toward the bid for $7.52M against open interest of 5,469. With FCX at 70.98 those 38 calls are 98.6% intrinsic — a delta-one stock substitute, not a bet — so this is a holder cashing part of a deep in-the-money position and re-striking twenty percent out and a year further on, with about $1M coming out net. That is a holder locking gains and staying involved, which caps nothing and shorts nothing; it is not new money arriving in copper while the strong-dollar block is on. FCX ZONE GAP: no zone row on the Monday sheet.

TAPE FCX 09:37:12 · Jan-2028 85C 5,128 toward the ask at 12.66 ($6.49M), open interest 4,355 · Jan-15-2027 38C 1,138 + 1,108 toward the bid at 33.44 to 33.45 ($7.52M), open interest 5,469

S&P September 25 5,900 puts — pennies in dollars, the dearest contract on the board in volatility

14,775 lots bought at the ask at 0.15, open interest under 1,500, about $220K. These showed on the low-dollar-price panel. They are struck twenty-three percent below the index and print at 60 implied against 9.3 at the money — more than six times the body. Structure checked: three single-leg prints, no matched call leg, all time value — an outright wing, not part of a box. A low dollar price is not a low cost: this is the contract where the seller has the better of it, and it is the reason the hedge at the end of this edition owns a near-the-money body instead.

TAPE SPX Sep-25 5900P 9,775 + 2,500 + 2,500 at the ask at 0.15 between 14:14 and 14:56, implied 60.1 to 60.7

Scorecard — grading the prior report

The Thursday edition, “Yields Blink”, was right on the pin and the banks and wrong on the durability of its own headline. Grade: C-plus.

The week ahead — what to watch

Bottom line

The week went nowhere on the index, narrow underneath, and the loudest number on the screen was an accounting event at the closing bell. The two-day lift was expiry mechanics on a washed-out mood; that support expired with it. What is left is a bond that could not hold its line for two sessions, banks and small caps that never bounced, crypto and memory carrying the tape on a two-session sample, a front-week volatility priced for nothing, and the market’s own protection dated to the next two Fridays. No index direction is licensed until the event card clears Monday. Read single stocks from price on volume, treat the accumulation ladders as quoted rather than proven until they are rebuilt without the closing print, and keep structures small, dated and defined.

Top Trades to Follow

CALENDAR BABA Sep-25 / Oct-16 115 call calendar — the one liquid name on the tape where the front is the rich contract: front-week implied 50.2 against 43.7 in October, front-over-back 1.15x, with spot up and implied volatility up on the day; sell the weekend headline premium, own the October body near the money; modelled debit about 2.35, maximum loss about 235 dollars; owned into the September 24 window, retired on a close under 23.54 in the China internet fund (KWEB) or at the front expiry.
DIAGONAL QQQ Oct-16 715 put / short Oct-02 690 put diagonal — own the fairly priced October 16 body (16.7 implied, put wing 1.18x its at-the-money) and sell the October 2 wing where the skew ratio is 1.58x; modelled debit about 7.50, maximum loss about 750 dollars; armed, not entered — it waits for the event card to clear Monday and for a rejection between 722.61 and 725, preferred September 24 to 30; owned into the payrolls window, void on a close over 725, retired into a flush toward 705 to 711.
TAIL SPY January 720 put tail, carried — struck at about 1.18x skew; its retire condition is an S&P close over 7,641 with the long bond over 81.35 in the same week, and the bond failed; carried into the late-September-to-October window; not replaced with a front-week wing priced at 2.12x its at-the-money; maximum loss the premium paid.
CALL TLT Oct-16 81 call, the surviving back leg of last week’s calendar — October implied 11.8 against 9.9 in the front week, a term ratio of 0.84x, and the one part of the surface that stayed firm; owned into the October window, retired on a close under 79.74, confirmed only by a weekly close over 82.50; maximum loss the remaining premium.
HOLD MSTR Oct / Nov call spreads: harvest strength, open nothing new — front-week implied 70.6 against 68.6 in October, a flat 1.03x term ratio with spot up and implied up together; after a plus 16% session the measured ten-session path includes a median dip near 12%; long call spreads are harvested into strength and no call is sold against it; re-entry window after a pullback toward the weekly lower at 140.97, retired on a close under the monthly middle at 132.94.

Sources

Expected moves & levels: Silva daily board, the ZONE DOCUMENT and the Zone Visual for 09/21 (zone document ingested this session; the Zone Visual read page by page — its room percentages come off an earlier mark than the document’s close), weekly board 09/21 to 09/25, the expiring weekly 09/14 to 09/18, monthly September and quarterly Q3 boards; Asher and census implied-volatility brackets; the FOM sentiment gauge for 09/18 (43.2, logged this session). Tradytics: Live Options Flow and Darkpool Market Summary end-of-day files for 09/18; the five prior end-of-day options files 09/11 to 09/17, replayed through the same decomposition; the 26-page dashboard capture for 09/18, every page read (the sector-flow line chart on page 11 reads as stacked and is not cited; no QQQ zero-day panels in the capture). Recon pipeline: the 2026-09-18 census, 652 per-ticker reports, and all 652 timestamped dark-pool print files behind the closing-print measurement; five names cited only as index events have no census file (ILMN, TAP, DINO, YPF, CAR). Ledgers: flow, dark-pool, monetization, setup-scan, expected-move and volatility-tracker ledgers for 09/18. Commentary (09/12 to 09/18): about 110 polished transcripts read by two reader passes, with the Maverick, Volsignals, SpotGamma and Silva passages checked directly against the transcripts; the Maverick’s five morning briefs; Tim Knight’s four files of 09/16 to 09/18. Timing: the updated September projection pair and its author’s 09/18 post. Web (index events only): the S&P Dow Jones Indices quarterly rebalance release of 09/04; an Investing.com piece on the SpaceX Nasdaq-100 float adjustment. Market data: daily closes 09/04 to 09/18. Framework: comprehensive_analysis_0918.md, daily_close_recap_0918.md, the prior edition, THESIS_LEDGER, TIMING_LEDGER. Not integrated: the Silva slide deck of 09/16 (its transcript was); no bond timing chart is on the shelf.