Daily Report — 09/11/26 · “Every Hedge On The Tape Is Dated October”

The week in one line, and the finding that matters

Four down closes, one up close, and a market that quietly moved every hedge it owns out of September and into October. The S&P finished the week at 7,655.50, down 0.82% from the 7,718.60 anchor, after losing its weekly floor on Thursday and reclaiming it on an in-line inflation print Friday. That is the visible story and it is the less important one. The structural story is on the calendar: the quad-witch expiry that carried a deep put book all through August now carries none, the week after it is where the new protection sits, and two of the largest option structures printed on Friday were hedges being rolled forward in the same second they were closed.

The bull case and the bear case for SPX this week are not arguing with each other. They are arguing about different evidence. The bear case is a rates case: the ten-year closed 4.975%, a new high for the year and the highest since the 2007–2008 area, the two-year at 4.59% sits 96 basis points above the effective funds rate, and a quarter-point hike is 86.5% priced for Wednesday. The bull case is a positioning case: breadth washed out to a roughly fifty-in-a-thousand extreme, dark pools bought technology all week, and the one bearish setup this framework can see in the options tape fires on precisely zero names out of three hundred and twenty-nine. Neither side gets to claim the tape. What both sides agree on is when.

Regime — a hike into a soft consumer, with oil driving the whole machine

The single most useful number of the week is a correlation, not a level. The twenty-day correlation between the ten-year yield and crude oil is 0.95. Oil ran 9.1% on the week and took rates with it; rates took the multiple with them. That chain — barrel to yield to price-to-earnings — is the whole regime, and it explains why a market with record earnings cannot get out of its own way. At the current ten-year the fair multiple works out near 11.4x, and the ten-year real yield at 2.42% now sits above the 2.35% breakeven, which is the condition that de-rates the highest-multiple names first. It has been doing exactly that: IGV, the enterprise-software fund, lost its lines on Monday and never took them back, with MSFT and NOW both closing under theirs.

The Fed is not the whole story either, but it is the near one. A quarter-point move to 3.75–4.00% is priced at 86.5% against 48% a month ago, and the two-year crossing above the funds rate is historically the market telling the Fed where it is going rather than the reverse. The consumer is the channel that matters if this goes wrong: preliminary September sentiment came in at 47.8 against a 51.0 consensus, zero consumer-discretionary members sit at 52-week highs while 21.3% sit at 52-week lows — the worst since April 2025 — and XLY closed below its monthly one-standard-deviation floor at 113.58. Retail sales land Wednesday morning, hours before the decision, with the prior print at -0.6% and consensus at +0.9%. That is the one number next week that can turn a policy story into a growth story.

Fear is present and capitulation is not. The sentiment gauge closed Friday at 40.5, up 7.9 on the day and still down 13.4 over five sessions — it entered the fear band at 32.6 on Thursday and vacated it in a single session. Either reading is a long way from the sub-15 prints that have marked contrarian bottoms and equally far from the above-80 readings that mark the other extreme, and a gauge that bounces with the tape rather than leading it is a neutral input rather than a bullish one. That one-session round trip is itself evidence for the mechanical reading of Friday: mood followed price, it did not cause it. The dollar was flat on the week, so the hard-asset block is not active; gold, silver and the miners all closed inside every band they have.

The four timeframes — nothing is beyond a band on the index, which is the point

For the first Friday in weeks the index owes nothing to any of its four bands, and that is precisely why the index is not the signal next week. The S&P closed mid-band inside Monday's daily band, a daily expected move of 7,608.48 to 7,705.48 with the two-deviation extension at 7,559.98 and 7,753.98; mid-band on the new weekly band 7,531.40 to 7,782.56, a tenth of a deviation below centre on the September monthly 7,428.73–7,943.55, and a quarter-deviation above centre on the quarter's 6,929.27–8,069.45. No index touched its quarter-to-date ceiling or floor, so there is no multi-index breach and no regime signal from the longest timeframe. The implied-volatility bracket agrees with the dealer bands on the week and the month and prices Monday tighter than they do, at 7,625.16–7,688.80.

The forward week is the tell inside those numbers. The weekly expected move widened 32% to 125.58 points while the daily contracted to 48.50, so the market has concentrated its entire risk budget on Wednesday's decision and Friday's expiry and is charging almost nothing for Monday. That has a direct consequence for how any view gets expressed, and it is in the trade section below.

Monday's forward zone set landed over the weekend and it is the fifth frame, not a footnote. The index zone reads 7,588.97 / 7,669.08 / 7,749.19 and the Friday close sits twelve points under its own zone mid — so the tape enters the Fed week fractionally heavy on the structure that matters most for a pin, with the zone high and the daily two-deviation line stacked within five points of each other at 7,749 to 7,754. QQQ is the exception and it matters: at 714.88 it is the only index closing above its zone mid 714.61, with a zone high of 723.07 sitting immediately above the 720 breakout level. Those two numbers bracket a three-point confirmation band, and semiconductors are what decides whether it clears.

Away from the index, five things sit outside a band and each one carries a different obligation. The long bond at 80.87 is below its weekly floor 81.35 and its quarterly floor 82.50 and hangs 0.67% above its monthly floor 80.33 — that is continuation, not reversion, because the regime input moved underneath it. The forward zone sharpens it into a shelf rather than open air: the zone low 80.40 sits within seven cents of the monthly floor, so two independent structures agree on a floor half a percent below the close, and the zone mid 81.90 sits just above the 81.35 line that reopens the yield gate. TLT has to clear both to turn, and it has a defined place to stop if it does not. Small caps spent four sessions under their old weekly floor at 291.16 and are the cleanest band-break reversion candidate on the board, where the measured base rate is a next-session bounce 89% of the time and a full re-entry inside a week 74% of the time. META at 648.03 is the only mega-cap above its monthly ceiling 618.08 — continuation by sponsorship, but a monthly-ceiling tag caps the next expiry rather than confirming it. Crude settled one tick under its weekly two-deviation ceiling at 100.18, which is a reversion lean inside an intact uptrend. And high-yield credit at 78.60 is not quite on its floor after all — Monday's zone puts the low at 78.35, so credit sits 0.32% above it rather than on it, with the zone mid 79.06 overhead. That is a softer statement than the Friday map supported and it is the honest one: a watch line approached, not a break, and the correction matters because the whole credit read rests on it.

TAPE · Dark pool: technology sector net +$5.1B on $39.08B of gross, the largest inflow of the session; financials -$8.2B, the largest outflow. Options: technology cumulative premium at a new month-to-date high near +$425M; financials the only sector with a negative per-name average at -$1.85M.

The calendar finding — three panels, one message

Every meaningful hedge that printed on Friday is dated October or later, and two of them were rolled out of September in the same second they were closed. This is the cycle's actual finding and it came from three places that do not talk to each other.

First, the expiration flow timeline. The September-18 index line — the quad-witch — ran to about -102 at its August trough as put protection piled in, and has now netted all the way back to +8. The protection that was built through August has been unwound. Meanwhile the September-25 line, the week after the expiry, has built steadily since 08-24 to -66 and is sitting flat at its low. The hedge did not disappear. It moved one week to the right.

Second, the dealer delta book. The mid-September cluster carries roughly +2.9B of call deltas against -6.4B of put deltas, a net short book near -3.5B — large, but smaller than the -5.2B the same panel showed two sessions earlier. The book is being reduced into the expiry, which is what a floor looks like while it lasts and what a vacuum looks like the moment it rolls off.

Third, the raw prints, which are the least ambiguous of the three. Financial-sector protection was rolled from the September-18 58-strike into the October-16 56-strike — about 107,625 contracts sold against 101,250 bought, at 14:19:34, paying two volatility points to extend. Medical-device protection was rolled from the October-16 52-strike into the January-2027 50-strike, about 62,000 against 64,000, at 12:20:58. On top of those, 50,000 October high-yield credit puts were bought at the offer, SQQQ, the three-times-inverse Nasdaq fund, saw October calls bought and October puts sold, twenty thousand October-21 volatility calls printed, and a January-2028 long-bond put structure went up in size. There is no September hedge left worth naming.

Put the three together and the market's own option book has an opinion about the shape of the month: supported into Friday, unprotected after it. That is the same window where the two competing timing maps disagree about direction — which makes the week of the 22nd the only window worth pre-positioning for, and it makes the expression October-dated by construction.

TAPE · Options: financials October-16 56-strike puts 101,250 bought against September-18 58-strike 107,625 sold, one timestamp; high-yield credit October-16 80-strike puts 50,000 at the offer for $7.5M at 4.78 implied against a 4.63 at-the-money; medical devices January-2027 50-strike 63,960 bought against October-16 52-strike 61,992 sold.

The panel walk — where the venues disagree

Zero-day flow: the index venue bought, the fund venue sold

The cleanest divergence on Friday's board was between two instruments tracking the same thing. The S&P index zero-day net flow ran positive all afternoon, holding near +$4M with two large green prints around 11:30 and 12:20. The S&P fund's zero-day net flow ran the other way, sliding to -$3M to -$4M from midday into the close. Same underlying, opposite sign. Cash-settled index paper is where institutions express same-day views because there is no assignment risk; the fund's zero-day book is where retail and hedging sit, and its single largest line was over a million 765-strike puts. The reading that survives is that the professional venue was buying the relief and the hedging venue was paying for it.

The strike map underneath was pure expiry mechanics. The index carried a positive-gamma spike near +$8.2B at the 7675 strike, sandwiched between roughly -$15B at 7685 above and -$9B at 7655 below, which pinned the close into the 7645–7675 pocket exactly where it landed. The fund's board was negative at every single strike with its deepest hole at 765. All of it expired at the bell, so none of it is a forward level — Monday opens with the pin gone and the forward flip line at 7,636 immediately underneath.

Delta exposure, flow map and the long-dated tails

Delta is being added back slowly, not aggressively. The market delta gauge printed its third and fourth consecutive green session, but the bars are small — a few tenths of a billion against the -1.5B red bar earlier in the month. That is a same-session amplifier and never a next-session compass, and at this size it is not even much of an amplifier. The flow map by expiration tells the more interesting story at the long end: the largest positive bucket on the board sits at a late-October expiry near +$34M, and the largest negative by a wide margin sits at June 2027 near -$63M, driven by call selling rather than put buying. Someone is financing 2027 by writing calls and spending the proceeds in October.

Top flow and where the volume changed

The two largest anomalies in volume change resolve in opposite directions once you look at what actually printed. SMH traded roughly 140,000 puts against a 26,000 average — a five-fold outlier and the single biggest deviation on the board. It is not a hedge. It is 93,671 contracts of a six-cent, five-day, 23%-out-of-the-money strike at roughly three times the at-the-money implied volatility: about $547K of premium inside a $52.3M name. Strip the structure and the residue is positive, and it flipped sign in the process. SMH was not shorted on Friday; it was decorated with a lottery ticket, and the lottery ticket was expensive in the only unit that counts.

MSTR was the other outlier — roughly 100,000 calls against a 35,000 average — and that one is real and coherent. Three September-18 call spreads printed inside ten minutes, struck 135/141, 136/142 and 138/143, for about 32,000 lots, with one leg at the bid and the other at the offer in the pairing that proves a genuine vertical rather than a coincidence of adjacent strikes. On top of it, a December-2028 130/90 put spread went up, roughly 2,800 against 3,640 lots. Near-term convexity owned, multi-year tail hedged, nothing sold naked against the stock. That is the same architecture this framework applies to the name, running at about three thousand times the size.

TAPE · Options: PWR +$34.94M readable at 81.1% of gross, the largest of the day; XLF +$10.20M at 38.5%; AMAT +$7.34M at 66.4%; on the sell side AMR -$22.14M at 100%, MRK -$20.01M at 28.2%, HWM -$14.24M at 99.5%. Dark pool: META +$2.98B at the offer, MU +$1.8B, AAPL +$1.9B, against NVDA -$3.12B and LRCX -$1.39B at the bid.

Unusual trades

The financial-sector put roll — the biggest structure of the day, and it inverts its own label

At 14:19:34 a single participant sold 107,625 September-18 58-strike puts on XLF and bought 101,250 October-16 56-strike puts. The aggregate residue for the fund reads as a +$10.20M buy at 38.5% of gross — well above the threshold where a residue becomes citable — and if you stopped there you would call it bullish positioning into a curve-steepening hike. The structure says the opposite of that: this is protection being closed at the expiry and re-established one month out and two strikes lower, and the implied volatility paid to extend rose from 14.28 to 16.37. Set it beside the -$8.2B the sector took on the dark tape and the picture is coherent — financials are the market's funding source this month, and whoever is long them intends to stay hedged through October.

TAPE · Structure-adjusted, this is one two-leg roll and not two directional prints: XLF Sep-18 58P 107,625 sold at 0.87 into 104,709 open interest (the closing leg); Oct-16 56P 101,250 bought at 0.61 into 16,029 open interest (the opening leg); sector dark-pool net -$8.2B, per-name average options premium -$1.85M.

PWR — one synthetic long, and a sixteen-session book pointing the other way

The largest readable options figure on the tape belongs to PWR, and it is not a campaign. Three clips at 11:09, 11:12 and 11:16 sold 2,790 December-18 760-strike puts to the bid at roughly $128.50 with the stock near 650 and open interest at zero — an opening synthetic long, about $35.85M, which is essentially the whole residue. Against that, the sixteen-session dark-pool ladder reads distribution with a cumulative net of -$443.92M, a falling three-day slope, and the ladder-versus-price contradiction flag firing on a session the stock rose 5.15%. One institution took a big synthetic long in the quality end of grid and power on a strong day while the multi-session book was distributing. Both of those are true and they belong in the same sentence.

TAPE · Structure: a deep in-the-money synthetic long, not a directional put sale — PWR Dec-18 760P 2,790 sold to bid across three clips, open interest 0, roughly $35.85M, extrinsic about 14% of the contract; census ladder DISTRIBUTION, cumulative dark-pool net -$443.92M, three-day slope FALLING -$50.86M, ladder-versus-price contradiction flagged.

Gold's 2027 straddle — $23.6M with no direction in it at all

At 11:42:08 a 3,000-lot September-2027 400-strike straddle printed on GLD, both legs at zero open interest, $14.59M of calls and $8.97M of puts at 24.29 and 24.81 implied. Matched strike, matched size, matched second: it carries no directional content whatsoever and would corrupt any naive aggregate it landed in. Somebody wants to own a year of gold volatility with no view on the sign. Separately and genuinely directional, 2,000 February-2027 415-strike calls were bought at the offer for $4.75M.

TAPE · Structure: matched legs, a straddle carrying zero directional content — GLD Sep-2027 400C 3,000 and 400P 3,000, both open interest 0, one timestamp, $23.56M combined; GLD Feb-2027 415C 2,000 at the offer, $4.75M at 24.45 implied.

BIDU — a dated synthetic long in Chinese internet, before the dark tape has turned

BIDU printed a +$4.09M readable residue at 65.6% of gross, and the structure behind it is 2,000 March-2027 110-strike puts sold to the bid for $4.56M — deep in the money, an opening synthetic long with a seventeen-month horizon. What makes it worth flagging is the contradiction underneath: the name's sixteen-session cumulative dark-pool position is at a new low with a falling slope, and the China internet fund never printed the 25.15 reclaim its reversal watch needs. The options venue has taken a dated long position before the cash tape has confirmed anything. That is a watch, not a trigger, and the trigger is the reclaim.

TAPE · Structure: a deep in-the-money synthetic long rather than a directional put sale — BIDU Mar-2027 110P 2,000 sold to bid, $4.56M, 44.74 implied, open interest 896; census ladder SUPPRESSED on a tag-versus-net conflict, three-day slope FALLING -$92.16M, cumulative at a NEW LOW.

The long-bond 2028 put structure — the rates thesis, expressed in size

At 11:06:06, 7,406 January-2028 80-strike puts were bought at the offer on TLT for $2.85M at 12.44 implied, matched leg-for-leg against 7,406 November-20 83-strike puts in the same second — a two-leg calendar structure rather than a naked outright — with a further 3,000 January-2028 78-strike puts behind it. With the fund at 81.19 intraday, that is a multi-year position for yields continuing higher, and it priced at roughly one and a half times the near-dated at-the-money volatility — a fair price for duration rather than a wing chase. It is the single clearest expression on the tape of the thesis that is de-rating the multiple.

TAPE · Structure: a two-leg calendar, matched size across expiries rather than a single outright — TLT Jan-2028 80P 7,406 at the offer, $2.85M, 12.44 implied, open interest 100,709, against Nov-20 83P 7,406; plus Jan-2028 78P 3,000; fund closed 80.87, below its weekly floor 81.35 and quarterly floor 82.50.

Single names — the mega-cap roster, priced against its bands

The mega-cap complex — AAPL, MSFT, GOOGL, GOOG, AMZN, TSLA, NVDA and META — split into three groups this week, and the split is more informative than any of the individual moves.

Bought on both venues. AAPL closed 332.27, up 1.75%, sitting 0.86% under its monthly ceiling 335.12, with $2.28B of dark-pool volume at the offer against $386.7M at the bid and call buying in the eighty-first percentile of its own history. MSFT is the more interesting of the pair: it closed 495.63, still under the 499.40 line it lost on Monday and 3.6% above the 478.14 level that would turn a de-rating into a problem — and its sixteen-session dark-pool book is at a cumulative new high with a rising slope. The cash tape is buying the de-rating while the price sits under its own line. That is the cleanest options-versus-price divergence in the mega-caps, and Monday's zone confirms exactly where it resolves: the zone mid is 498.76, within sixty-four cents of the 499.40 line the stock lost. Two independent structures naming the same reclaim is worth more than either alone, and the zone band 481.95 to 515.56 gives the trade both a stop and a first objective.

Sold on one venue, bought on the other. GOOGL closed 338.50 and GOOG 335.45, up 1.77% and 1.53%, and between them they were two of the five largest options buys of the day — and both printed large dark-pool sells, -$1.50B and -$1.35B at the bid. The options residue on the larger line is 21.3% of gross, under the threshold that makes a residue citable, so it is structure-dominated and carries no direction. One session of mega-cap selling on the dark tape is a description and never a bearish read; holders locking gains is the likelier explanation and it does not change the trend. The Alphabet split is noted and not resolved on one session. On the forward zone GOOGL at 338.50 closed just over its mid 337.62 and GOOG at 335.45 just over its 334.59 — both marginally constructive, neither decisive. AMZN at 256.78 sits under its zone mid 257.18 and TSLA at 365.44 above its 362.04; both closed inside every band with no readable residue and nothing to say beyond that.

The one that matters. NVDA closed 218.29, flat, and it is the single mega-cap where price, the multi-session book and the positional map all point the same way. The sixteen-session ladder reads distribution, the cumulative dark-pool net is -$6.19B with a three-day slope falling -$5.80B, and the close sits under the 221.45 weekly floor it lost. Its options residue is 1.7% of gross — structure-dominated, no direction in it — so the flag is the cash book and not the tape. It is explicitly not a short: no exhaustion tag, no break of the 207.68 zone floor, and no seller-based read has ever been a bearish setup in this framework. It is a name on the buy shelf whose book has not turned. Price says shelf; book says not yet. The discriminator just got sharper: Monday's zone mid is 221.88, within forty-three cents of the 221.45 weekly floor the stock lost, so the reclaim line is now named by two independent structures and sits 1.6% above the close. The zone low 208.41 is the mirror — it lands on the 207.68 trend floor, which means the whole NVDA question is bracketed by two places where the frames agree, and nothing in between is a signal.

Semiconductors, which are now the whole index. AMD closed 516.13, up 2.49%, with $1.04B of dark-pool volume at the offer and literally none at the bid, and holders writing calls into a 13% trailing run — sponsorship with a cap, not a top. INTC at 102.94, up 2.61%, carries a rising slope at a cumulative new high and twelve of sixteen bullish sessions. MU at 975.26 held the 952.03 line but is down 5.11% from Wednesday with both ladder warnings firing at once — the multi-session tag contradicts both the day's price action and its own slope, which under this framework's three ladder guards is a veto rather than a footnote. On the forward zone the cohort is split by position, not by direction: INTC sits highest at 102.94 against a mid of 96.55, AMD next at 516.13 over 487.47 with room to a 530.59 ceiling, while MU at 975.26 is sitting on its zone mid 973.65 almost to the dollar — dead centre, which is what an undecided name looks like when the ladder has already been vetoed. AVGO at 361.99 reclaimed 361.40 and closed fractionally under its zone mid 362.61, and its sold-call bucket is 43.3% far-out-of-the-money, which is a structure label and never a direction. SMH, AMAT, MRVL and SNDK carry no row on Monday's zone document, so those four are read on their iVol brackets instead (the implied-volatility outer band): SMH ZONE GAP, weekly iVol 545.61 to 591.45; AMAT ZONE GAP, weekly iVol 425.50 to 487.48; MRVL ZONE GAP, weekly iVol 216.42 to 255.78; SNDK ZONE GAP, weekly iVol 1,493.03 to 1,773.67. A missing zone row is never a missing band. SNDK at 1,633.35 is down 7.41% in two sessions from the level at which this report named it a top trade, its ladder suppressed and its slope falling — that idea is withdrawn below, not softened.

The rest of the roster, levelled. NFLX closed 77.40, up 1.83%, with a readable +$5.75M buy at 26.6% of gross — the only roster name with a citable residue — while sitting six-tenths of a deviation below centre on its monthly 75.30–86.80, the weakest monthly position on the board. SPCX at 151.21, up 2.04%, carries a cumulative dark-pool net of +$2.77B at a new high with a rising slope and eleven of sixteen bullish sessions, the strongest ladder of the roster. NFLX's zone is 74.32 / 79.02 / 83.71, so it closed under its mid on the weakest monthly position of the roster — the citable buy is being put to work into structural weakness, not strength. MSTR at 130.97 is sitting on its weekly floor 130.66 with momentum in the ninety-ninth percentile and the call-spread architecture described above; MSTR ZONE GAP, weekly iVol 118.78 to 143.16; SPCX ZONE GAP, weekly iVol 140.69 to 161.73 — neither carries a row on Monday's document, so both are read on those brackets. META at 648.03 is the only name above its monthly ceiling and it also sits well above its zone mid 622.19, with the zone high 676.00 the next structure overhead — that is the number the capped-upside read was missing. Its holders are already writing calls against a 6.12% trailing run: bullish-to-neutral, capped into the next expiry, and never a short.

TAPE · Dark pool: MSFT cumulative +$8.50B at a NEW HIGH, three-day slope RISING +$5.27B; NVDA cumulative -$6.19B, slope FALLING -$5.80B; META cumulative +$13.14B at a NEW HIGH. Options: NFLX +$5.75M at 26.6% of gross, the roster's only citable residue; AMD sold-call bucket 72.6% monetizing against a +13.15% trailing run; AVGO 43.3% far-out-of-the-money writing, a structure tag.

Sector rotation, built from the constituents up

Two sectors moved on both venues in the same direction this week, and they moved opposite ways. Technology took +$5.1B of dark-pool net on the largest gross inflow of the day and pushed its cumulative options premium to a new month-to-date high near +$425M — above the early-September peak — while the index itself fell on the week. Underneath that, the buying is everywhere except in the two names that led the last leg: AAPL, MSFT, MU, AMD, INTC, TSM and AMAT all took net dark-pool buying, while NVDA and LRCX were sold. A sector being accumulated everywhere except in its recent leaders is a rotation inside a rotation, and it is why the semiconductor fund is now the only instrument still amber rather than red on the five-day direction board that tracks these things. If semiconductors go red, the Nasdaq has nothing holding it.

Technology's zone position backs the flow: XLK at 187.67 closed above its mid 186.44, the only major sector fund doing so alongside communication services at 112.60 over 112.19. Financials went the other way on both venues and did it every single session: XLF at 57.25 under its zone mid 57.60, -$8.2B of dark-pool net, the only negative per-name options average on the board at -$1.85M, and a cumulative sector line making a new low daily. GS, MA and AIG were sold on the dark tape against JPM bought. The fund-level options number looks like a buy and is a hedge roll, as above. Financials are where this market raises cash.

Communication services is the concentration trade: the highest per-name average premium of any sector at +$4.6M, and it is three tickers deep — META, GOOG and GOOGL. Consumer staples took the second-largest dark-pool inflow at +$1.9B with PG bought at the offer on a 65% volume expansion, and alongside it the municipal-bond fund printed a genuine multi-print accumulation on a 53% volume expansion while the Treasury-bill fund took $578M across 39 blocks. Money moved to the front end, to tax-exempt paper and into staples on the same day the long bond made new lows. That is a duration barbell and it is the most under-discussed print of the session.

Healthcare is the week's one genuine puzzle. It was the worst sector on price at -3.55% with biotech down 4.65%, and its cumulative options premium swung roughly +$210M in two sessions from deeply negative to positive — the largest rate-of-change on the board. At the same time the IHI medical-device hedge was rolled a quarter further out, and XLV itself closed barely above its monthly floor. Options green, price red, protection extended — and on the zone, XLV at 165.36 sits between its low 162.97 and its mid 169.38 while biotech at 156.20 is only 1.3% above its own 154.25 floor. That is a stabilization watch and it needs a second confirming session before it is anything more. Energy led the week at +1.69% with the barrel over its continuation line, but XLE's own option structure is a holder overwriting — long-dated in-the-money calls sold and January puts sold — and the residue falls just under the citable threshold. The barrel leads, the equities follow, and nobody is paying up for the equities.

TAPE · Dark pool by sector: technology +$5.1B, staples +$1.9B, industrials +$550M, communication services +$450M, against financials -$8.2B, consumer cyclical -$3.7B, materials -$600M. Census screen: 652 files swept, 489 flagged, 94 carrying the ladder-versus-price contradiction veto.

Breadth — why it bounced, and what it has not repaired

Friday's rally was a mechanical consequence of Thursday's extreme, and the internals have not repaired. The share of S&P members above their twenty-day average closed Thursday near 18% and bounced to 25.60 on Friday, a 42% one-day move; the historically extreme threshold is 15% or lower, which prints roughly fifty times in a thousand sessions. Beneath that, 39.20% sit above their fifty-day and 59.00% above their two-hundred-day, so the primary trend is intact and the near-term damage is real. The S&P bullish-percent index is at 43.20 having been near 72 five weeks ago, with a fourteen-day momentum reading of 18.95; the NYSE reading is 46.46 at 17.29; industrials 31.65 at 22.88; transports 25.00 at 25.74; discretionary 18.75, down 10% on the day, at 20.83. Four separate breadth gauges are simultaneously oversold while the index sits 1.9% off its high. That combination resolves either into a successful bottom test or a failed one, and the distinction is the single most important thing to watch next week.

The internal baskets say the repair has not started. The AI basket covered 33 of 33 names with fifteen distributing against five accumulating and a falling multi-day slope on ten, and it raised a state alarm for the second consecutive night — a descriptive multi-session state, never a single-day exit signal. The speculative basket was worse: fourteen distributing against one accumulating, with three names at a cumulative dark-pool low. A market whose index rose 0.84% while both internal baskets deteriorated is a market where the average share did not participate.

TAPE · Breadth: 25.60% above the 20-day, 39.20% above the 50-day, 59.00% above the 200-day; bullish-percent S&P 43.20, discretionary 18.75. Baskets: AI 15 distributing / 5 accumulating, state alarm night 2; speculative 14 distributing / 1 accumulating.

Timing — the September map flipped its sign, and the clock did not move

The cycle analyst who publishes the September projection posted an update on Friday saying he had called the orientation incorrectly and that the inverted version is the live one. Timing charts give timing and shape, never price targets, so what follows is inflections only and every level in this report comes from the expected-move boards instead.

The September 6th post named a high around the 9th and a low around the 15th. The Friday update keeps both dates and reverses the sign, pointing to a high around the 15th. The dates never moved because the two charts are exact mirrors of one another, so the calendar of turns is identical in either orientation and choosing between them is a pure sign decision. Graded against the tape rather than against the author: the standard version needed a local high around the 9th and there was none — the high was the 3rd and 4th, and the 9th was the third of four consecutive down closes — so that orientation is dead. The inverted version needed a low around the 9th and the leg's low closed on the 10th with the turn arriving on the 11th, which is a hit inside one session. This report backed the standard orientation two sessions ago and that call is overturned; the miss is logged below.

What the live orientation projects forward, weekend boundaries mapped to trading sessions: a peak Monday the 14th or Tuesday the 15th; a high plateau rather than a spike, holding through the Friday expiry or the Monday after; a decline from the 22nd through the 24th; the month's low at the Friday 25th close or the Monday 28th session; a sharp one-to-two-session bounce around the 28th; and lower into October 1st with payrolls on the 2nd. The plateau detail is the one that matters: the chart does not draw a reversal on the 15th, it draws a rolling top that survives the expiry and breaks after it.

That collides head-on with this framework's own September map, and the collision is precise. Both agree the trough printed on the 9th and 10th. Both agree the Friday expiry is the pivot. They are 180 degrees apart on what comes out of it: the framework's base path calls the 22nd-to-25th window the month's peak, and the live external orientation calls the 25th-to-28th window the month's low. Same window, opposite sign, and the market's own hedge book — which moved out of that expiry and into the week after it — says something happens there regardless of which sign wins.

Calibration matters more than the call. On the August charts this model's dated turns landed and its signs did not: the late-July trough hit, the August 5th top hit exactly, the August 7th fade missed; the August 24th-25th up-leg confirmed, the August 25th month-high missed and the late-August trough missed. The dates land; the sign is the failure mode — which is precisely what the author said about himself on Friday. Use the clock, not the sign. On that basis the path weights move on the tape rather than on the flip: the base path down to roughly 35%, the inverted path up to roughly 40%, the squeeze path at roughly 25%.

Two independent voices land on the same windows. A widely-followed volatility desk called for a low in the 8th-to-16th window with volatility expanding first and then a squeeze into the Fed, and drew a hard line at holding no short past the 28th — his squeeze into the Fed is the 15th peak, and his alternative low before quarter-end is the 25th-to-28th trough. Separately, the desk that publishes the expected-move boards spent Friday's report on a single observation: the weekly volatility bands have compressed to a width of 7.864, a reading that has historically preceded expansion rather than continuation, and he expects a significant move between now and October without naming its direction. His weekly momentum shows a bearish divergence and a fresh downside crossover; his breadth work says the bounce was mechanical. He is openly two-sided and says so. So is the tape.

Tier moves

• Long bond to CONFIRMED DOWNTREND. Four sessions below its weekly floor 81.35 and its quarterly floor 82.50, with the ten-year at a year high and the hike near-locked. Earned.
• Financials to CONFIRMED UNDERPERFORMER. Both venues, every session of the week, and the protection just moved to October. Earned.
• Small caps to CONFIRMED UNDERPERFORMER and simultaneously a REVERSION CANDIDATE. Four sessions under the old weekly floor 291.16 with a genuine multi-print distribution campaign on a 152% volume expansion — and a band-break reversion setup that pays 89% of the time next session. Both are true; they resolve on 291.25.
• MRK and HWM to CONFIRMED DISTRIBUTION. Both at cumulative lows with readable directional selling at 28.2% and 99.5% of gross — the only two names where price, the multi-session book and the options residue all agree down. But position argues against chasing either: MRK closed 143.93 against a zone of 142.16 / 148.43 / 154.69, only 1.2% above its floor, so the confirmed distribution is arriving at support rather than through it. HWM ZONE GAP — no row on Monday's document; its weekly iVol bracket is 218.54 to 240.68 and the close 229.61 sits mid-bracket. Confirmed as a state, not as an entry.
• AMAT to CONFIRMED ACCUMULATION. Thirteen of sixteen bullish sessions, +$3.98B cumulative, a rising slope and a readable buy at 66.4% of gross. The cleanest two-venue positive on the board.
• Semiconductors to STABILIZATION WATCH from leader. Two up sessions bracketing two down, with AMAT confirmed and LRCX decaying inside the same cohort. A cohort that splits internally is a watch.
• SNDK: top-trade idea WITHDRAWN. Down 7.41% from where it was named, ladder suppressed, slope falling.

Convergence, and the honest probability

Six bearish themes against one bullish, with three neutral — and one up session against four down. The bearish side: policy with a hike 86.5% priced, rates with the ten-year at a year high and the real yield above the breakeven, the oil cost channel, financials selling on both venues, consumer-discretionary breadth at its worst since April 2025, and a dealer calendar that has moved its protection past the expiry. The bullish side is one theme and it is large: technology bought on both venues with its cumulative premium at a month-to-date high. The neutrals: sentiment, which entered the fear band at 32.6 on Thursday and was back to 40.5 and neutral by Friday's close — a one-session round trip that followed price rather than leading it; index breadth washed out, which cuts both ways; and the setup board, which needs reading carefully.

That last one changed this week. The exhaustion pattern — a vertical run with call buyers chasing and holders selling calls into them — graduated on Friday from an advisory flag to a validated tactical bearish setup on a forty-nine-date forward grade, where 35% of tagged names fell 10% or more inside ten sessions against a 13% base. It is now the one bearish setup this framework will act on, and it fires on zero names out of three hundred and twenty-nine. The same grading exercise found that the sponsorship tag underperformed the untagged universe — on names like AAPL and META the overwriting caps the name — so sixty-eight sponsorship tags are a description of state, not a bullish input, and this report counts them as neutral.

Net that out and the direction is stated and the conviction is not: the weight of evidence leans lower into the back half of the month, and the price has not confirmed it. The base path holds roughly 35%, the inverted path roughly 40%, the squeeze roughly 25%, and those are judgment tiers rather than model output. What converts the lean into a position is a failed retest, and what kills it is a close through the weekly ceiling.

The decisive causal read

Friday was a volatility event, not a demand event. Core inflation came in a tenth hot month-over-month and in line year-over-year, which removed the tail risk of a bigger move without changing the policy path; volatility fell 11.2% on the day, and in a market where the index sat just above its gamma pivot, falling volatility mechanically pushes dealers to buy. That is the entire move. It stopped 2.66 points under the daily ceiling, it was carried by three communication-services names and the AI-hardware read-through from a backlog headline, and eleven of the tracked instruments still closed under their weekly floors against five above. Meanwhile the name that actually reported gave back a seven-point opening gap and closed down 1.74% — a beat sold, which is the most honest single data point of the session.

Forward path — the week of the 14th, session by session

Monday is priced quiet, Wednesday is priced loud, and the window that matters is after Friday. The implied-volatility bracket prices Monday tighter than the dealer bands do, which is unusual and says the market expects the expiry week to start with a drift. The plateau shape both timing maps now share says the same thing.

• Monday 09-14. The reversion arm re-opens because no top-tier macro print sits inside the next eighteen hours — though note the volatility leg of that arm is marginal, not clean: the gauge closed 15.84 against a zone mid of 15.72, fractionally the wrong side, and the setup qualifies on the index sitting above its gamma pivot rather than on both legs. Small caps are the live setup: a close back over 291.25 triggers, graded Tuesday, with the zone mid 293.07 the first real objective above it; a close under 286.53 kills it, and the zone low 285.14 sits immediately beneath, so the invalidation is a 1.4-point cluster rather than a single line. The pin from Friday's expiry is gone and the forward pivot at 7,636 sits directly under the close, with the index zone mid 7,669.08 twelve points above it — that narrow band is Monday's whole range argument.
• Tuesday 09-15. The live timing orientation's first genuinely out-of-sample claim since the flip — a local high Monday or Tuesday with Wednesday lower. It grades cleanly, and a tape still making higher closes into Thursday kills the orientation.
• Wednesday 09-16. Retail sales at 07:30 Central with the prior at -0.6% and consensus at +0.9%, then the decision at 13:00. Three of the five Fed decision days this year produced larger-range sessions that closed at the lows; expect range expansion regardless of sign. A miss on retail sales hours before a hike is the specific combination that turns a policy story into a growth story.
• Friday 09-18. Quad-witch. The net-short book rolls off and, on the plateau shape, the expiry settles near the top of the range. That is mechanically friendly to every capped and short-call-spread structure and unfriendly to long call spreads struck above spot.
• The week of 09-22 to 09-28. The binary. The framework's base path says peak; the live external orientation says the month's low. The hedge book has already voted that something happens there. Quarter-end and the collar roll land on the 30th, payrolls on October 2nd.

Key levels

Up: 7,669.08 zone mid first — the close is under it — then 7,705.48 daily ceiling, then 7,749.19 zone high and 7,753.98 daily two-deviation stacked five points apart, then 7,782.56 weekly ceiling and 7,797.48 on the implied-volatility bracket, then 7,800 net-gamma call wall and the descending channel rail off the 7,816.70 swing high, then 7,890 collar short call and 7,943.55 monthly ceiling under the 8,069.45 quarterly line.

Down: 7,636 forward gamma pivot, then 7,608.48 daily floor and 7,588.97 zone low, then 7,559.98 daily two-deviation, then 7,531.40 weekly floor with 7,516.48 on the implied bracket and 7,500 net-gamma put wall — a structure level from dealer inventory, not a demand wall, and equal call and put open interest at a strike is box financing rather than support, then 7,428.73 monthly floor — a one-deviation monthly event and the honest size of the bear case — and 7,405.82 weekly two-deviation.

Monday's zone document and zone visual are both on disk and parsed — 179 symbols, with 82 of 82 census closes reconciling inside one percent. The index triple is 7,588.97 / 7,669.08 / 7,749.19 and every zone level in this edition is that forward set, not Friday's. The range and trend columns were refused by the parser for 37 symbols including the index itself, which is a declared gap on those two columns only and not on the triple. Fourteen names carry no row at all — SMH, AMAT, MSTR, SPCX, PWR, VST, CEG, DELL, SNDK, MRVL, HWM, BIDU, COIN and IGV are each flagged in place as a zone gap and read on their expected-move bands.

Cross-asset triggers: TLT reclaiming 81.35 opens the yield-turn gate for the grid-and-power names PWR, VST and CEG — all three of which are zone gaps on Monday's document and read on bands alone — with the zone mid 81.90 the confirmation above it and 82.50 the quarterly reclaim; the floor beneath is a two-structure shelf at 80.40 zone low and 80.33 monthly. Crude's continuation line is 95.70 and its weekly two-deviation ceiling 100.18. Small caps trigger at 291.25 toward the 293.07 zone mid and die at the 286.53 to 285.14 cluster. QQQ's breakout is bracketed by 720 and its zone high 723.07, weekly ceiling 730.30, floor 699.46, zone mid 714.61 just under the close. Credit's floor is 78.35 with the zone mid 79.06 the reclaim. XLY's downside objective is its zone low 110.60; META's capped upside is its zone high 676.00; NVDA's reclaim is the doubled 221.45 to 221.88 line.

Scorecard — grading the 09/09 edition

Grade: C-minus. Two of five calls worked and both of the winners were cautions rather than positions. The one big directional long lost 3.2% across the basket in two sessions, and the rates call was wrong in direction. The report was long the thing that fell and cautious on the things that went nowhere, which is an inversion worth naming rather than smoothing.

• Memory long — MISS. SNDK was named at 1,764.17 on a 17.8% residue share and closed 1,633.35, down 7.41%. A residue at 17.8% of gross is below the threshold at which a residue may carry a directional label at all, and labelling it anyway is the specific error. Withdrawn, not softened.
• Semiconductor cohort long — MISS. AMD 521.09 to 516.13, MU 1,027.77 to 975.26, INTC 106.24 to 102.94, MRVL 235.01 to 236.10. Basket roughly -3.2% over two sessions. The cohort call was right about leadership and wrong about timing, which in a two-session horizon is simply wrong.
• NVDA distributor flag — HIT. Called as a census flag and explicitly not a short: 223.67 to 218.29, down 2.41%, with the ladder still distributing and the slope still falling. The framing held and so did the restraint.
• META sponsored-but-extended, do-not-chase — HIT. 653.69 to 648.03, down 0.87%. A name that goes sideways is exactly what a do-not-chase predicts.
• Rates and volatility hedge — MISS. The long-bond September 83-strike call was framed as a bet the yield spike was near spent; the fund went 81.73 to 80.87 and the yield made a new high for the year. The volatility October 35-strike call lost with volatility down 11.2%. The report's own caveat — that chasing the wing after a move pays nothing — was correct even though the named contracts were not.
• Timing orientation — MISS, and it is the one that stings. The 09/09 edition read the standard version of the September projection as "the one tracking." The author reversed his own orientation two sessions later and the tape never made the high the standard version needed. Logged.

What the misses have in common. Four of them were calls made on a single session of evidence — a two-day residue, a one-day cohort bounce, a one-chart orientation. The two hits were both multi-session reads that declined to act. That is the lesson this week keeps teaching from different directions, and it is why the tier table above promotes nothing on a single day.

Bottom line

The market spent the week telling you when, not which way. Price did almost nothing on net; the hedge book moved a month. Every structure worth naming on Friday — the financial-sector roll, the medical-device roll, the credit puts, the inverse-Nasdaq calls, the volatility calls, the long-bond puts — is dated past the expiry, and the expiration flow line for the week after the quad-witch is sitting flat at its low while the expiry itself has netted back to positive. Two independent timing maps and one widely-followed volatility desk all mark the same late-September window, and they disagree about the sign.

Two things landed after the Friday close and both cut the same way. The sentiment gauge had entered the fear band at 32.6 on Thursday and was back at 40.5 by Friday — a one-session round trip, mood following price rather than leading it, which is corroboration for the mechanical reading rather than a contrarian bottom signal. And Monday's forward zone set puts the index twelve points under its own mid at 7,669.08, with QQQ the only index above its mid. The market goes into Fed week fractionally heavy on the structure that governs a pin, with the one index that is holding held up by semiconductors.

Into that, the evidence leans lower and the price has not confirmed. TLT and the rates complex behind it are the bear case and they are doing real damage to the multiple; the consumer is deteriorating into a hike; and breadth is oversold on four separate gauges with the index less than two percent off its high, which is the setup for either a successful bottom test or a failed one. Against that, technology is being bought on both venues with its premium at a month-to-date high, the one validated bearish setup in this framework fires on nothing, and volatility has just been crushed. The honest stance is to own the date and not the direction: buy time past the expiry while the term structure is giving it away, keep every expression defined-risk, and let Tuesday's turn and Wednesday's print pick the sign.

Sources

Expected moves and zones: EM_LEDGER.json; IVOL_LEDGER.json; EXPECTED_MOVES/DAILY/daily expected moves 0914.png; TV_FOM_DEM_Data_2026-09-14.txt; daily expected moves - ZONE DOCUMENT 0911.pdf; daily expected moves - Zone Visual 0911.pdf; Daily expected moves - zones 0911.png; Daily expected moves - range & trend 0911.png; EXPECTED_MOVES/WEEKLY/weekly expected moves - 0914 to 0918.png; TV_FOM_WEM_Data_2026-09-14.txt; EXPECTED_MOVES/MONTHLY/monthly expected moves September 2026.png; EXPECTED_MOVES/QUARTERLY/quarterly expected moves July to September 2026.png; JPM Collar levels Q3 2026.png; FOM sentiment index 0910.pdf; sentiment_index_tracker.md.

Dashboards and raw exports: OPTIONS_FLOW/options dashboard 0911.pdf — all 22 pages rendered and read individually as ANALYSIS_OUTPUT/dashboard_panels_0911/opt-01.png through opt-22.png; OPTIONS_FLOW/Live Options Flow - 0911_1615.csv; DARKPOOL/Darkpool Market Summary - 0911_1615.csv. The darkpool dashboard is retired by standing operator decision and is not a gap; its one load-bearing panel is rebuilt from the constituents.

Gated ledgers and scans: FLOW_LEDGER.json; DARKPOOL_LEDGER.json; MONETIZATION_LEDGER.json; SETUP_SCAN_LEDGER.json; EVENT_REGIME.json; REVERSION_BUY_STATE.json; GATE_LEDGER.md; THESIS_LEDGER.jsonl; ANALYSIS_OUTPUT/TIMING_LEDGER.md; ANALYSIS_OUTPUT/census_screen_2026-09-11.md; ANALYSIS_OUTPUT/BASKET_MONITOR/2026-09-11/digest.md; ANALYSIS_OUTPUT/BASKET_MONITOR/SPEC/2026-09-11/digest.md; ANALYSIS_OUTPUT/venue_call_sheet_0911.json.

Census, every name individually: recon_data/2026-09-11/wl1/analysis_results/ticker_reports/ — 68 per-ticker files covering SPY, QQQ, IWM, NVDA, MU, SNDK, AMD, TSM, INTC, MRVL, META, MSFT, GOOGL, GOOG, AMZN, AAPL, TSLA, AVGO, ORCL, ADBE, DELL, HPE, ANET, CRDO, VRT, PWR, AMAT, LRCX, KLAC, SMH, NFLX, MSTR, IBIT, SPCX, DRAM, COIN, HOOD, MARA, XLE, XLF, XLV, XLY, XLI, XBI, KRE, IGV, GLD, SLV, GDX, TLT, HYG, USO, XOM, MRK, HWM, JPM, PG, SNOW, NBIS, SKHY, BIDU, KWEB, VST, CEG, SMR, LEU, EWY and IHI. KHC, MUB, TENB and ETHA have no census file and are declared data gaps — no single-name claim is made on them.

Commentary and briefs, read in full: FOM - Mike Silva 0911 commentary - The Spring Is About Fully Loaded_polished.md together with mike silva - fom - Stock Market Report slide deck - 9_11_2026.pdf, all 44 pages rendered to ANALYSIS_OUTPUT/silva_deck_0911/ — the transcript and the deck are one input and both halves were read; Click capital 0911 commentary - THIS Isn't Going To End Well_polished.md; James InvestAnswers 0911 commentary_polished.md; Volsignals 0911 commentary - TGIF - Today's CPI + Live Education_polished.md; MAV 0910 commentary - Grab Your Piña Colada_polished.md; Geeks of Finance 0910 commentary_polished.md; Arete Trading 0910 commentary_polished.md; Morning briefs/FOM briefs/2026-09-11/fom_gamma_0911.json; Morning briefs/Mav briefs/2026-09-11/mav_brief_0911.md.

Timing: TIMING/Savino September 2026 SPX projection - 0911 update.png and its inverse twin, with the 09-06 originals in TIMING/archive/; ANALYSIS_OUTPUT/AN_september_2026_projection_0906.md; TIMING_MODEL_PROTOCOL.md.

Prior-cycle artifacts: ANALYSIS_OUTPUT/daily_close_recap_0908.md, _0909.md, _0910.md, _0911.md; ANALYSIS_OUTPUT/comprehensive_analysis_0909.md; daily_report_0909.html; OPERATOR_PORTFOLIO_ROBINHOOD.json; OPERATOR_PORTFOLIO_FIDELITY_IRA.json; regime_snapshot.md.

No URLs were fetched and no web search was run for this edition. Every figure traces to a file on disk.