DAILY REPORT · FRIDAY 10/30 CLOSE · THE LAST SESSION OF OCTOBER · INTC -28.6% ON ITS GUIDE AND RDDT -29.2% · QQQ CLOSED THE MONTH UNDER ITS OCTOBER FLOOR · THE S&P 500 INSIDE EVERY SWING BAND · THE EQUAL WEIGHT +4.95% ON THE MONTH AGAINST THE AI BASKET -23.6% · WHAT OCTOBER DID, WHO SOLD, WHAT WE SAW LATE AND WHICH ANALYSTS HAD IT

Daily Report — 10/30/26 · “October Broke the AI Trade, Not the Market — QQQ Closed Under Its Monthly Floor on Intel and Amazon While the Equal Weight Finished Up 5%”

The S&P 500 closed Friday at 7,580.07, down 1.82%, under its daily band (daily lower 7,670.07) and inside its weekly band (7,471.07 to 7,776.29), its October band (7,363.93 to 7,939.15) and its quarterly band (7,089.48 to 8,213.60). QQQ closed 690.49, down 3.47%, 11.60 under its October floor of 702.09, on the INTC and AMZN reports: INTC fell 28.58% on its guide, and 32 of our 33 AI names fell. On the month SPY lost 0.91% and QQQ 6.66% while the equal-weight RSP rose 4.95% and nine of eleven sector funds rose: a rotation with a crash inside it. This edition carries Thursday as well, measures whether buying puts was the only trade (half right: AI puts 2.50x, shelter calls 4.37x), who sold (hedge funds at a record pace, during the fall; the clean sellers were the issuers), what we saw late (the price said so on 10/8, our bottom line on 10/14) and which analysts had it. Mike Silva’s 43-page deck is read page by page. Bearish on the AI complex, no index direction, and our event card rules out new index-direction trades through Tuesday.

Amended Saturday 10/31. After this edition was published we changed the group-call rule: a call that goes quiet no longer starts a cooldown, so Crypto & Digital Assets and Copper & Base Metals fire on Friday’s close. Five bearish group calls stand, not three, no cooldown runs to Wednesday 11/04, and the AI basket, one input short, can fire from Monday. Across 44 quiet exits a group fell no more than its peers on the same dates; Friday’s 6.35% was one session, when all 50 groups fell 3.36% on average and the calls still standing 7.63%. One figure is corrected: 32 of 33 AI names fell on Friday, not 33 of 33.

Friday in one line — a company-event break to end the month, with rates, the dollar and credit standing aside

**The S&P 500 (SPX) closed at 7,580.07, down 1.82%, under its daily 2-sigma floor (7,619.75) and inside its weekly band (7,471.07 to 7,776.29), its October band (7,363.93 to 7,939.15) and its quarterly band (7,089.48 to 8,213.60): a bad Friday, not a broken month for the index. QQQ closed at 690.49, down 3.47%, 11.60 under its October floor of 702.09: the one major index to end the month through a monthly band, on the last close of the month.** That break is a continuation on company events, the INTC and AMZN reports, and we say so because the usual suspects did not move: the 10-year yield fell 0.4 of a basis point to 4.933%, the dollar fell 0.08, and credit gave up 0.19% in HYG. 212 of the 651 stocks and funds we track rose and 437 fell, median -1.41%, with 175 names down 5% or more. The equal-weight RSP lost only 0.79%, because XLE, XLV, XLU, XLB and XLRE all closed at their highest closes of the month while the chips fell.

How the day ran. SPX opened at 7,631.07, already under the daily floor, on INTC’s guide and AMZN’s miss. By 09:15 Central our 33-stock AI basket was down 7.9% against SPY’s 1.55%, and the $6.3B Binance hack headline hit COIN, MSTR and HOOD. By 10:15 AMZN, TSLA, GOOGL and META had recovered while SMH, NVDA, TSM, AVGO and MRVL sat at session lows. By 12:45 that recovery had failed everywhere but AMZN. The second leg came with a Kioxia NAND price headline: the first -1,000 NYSE tick printed at 13:22 Central, VIX touched 23.93, SPY 750.36, SPX 7,527.56 and QQQ 683.45, the last within a tenth of an expected move of its weekly floor 681.44. The close came back 52.51 points in SPX and 7.04 in QQQ. VIX closed 19.42. The AI basket closed down 8.16% with 32 of its 33 names lower, the second-worst day of the 54 sessions on our monitor.

What the break licenses for Monday. The SPX close is a weak reversion setup, graded by a Monday close over 7,580.07: in the current Fed-tightening regime 33 of 48 downside index-band breaks reverted the next session. QQQ’s touch of its weekly floor is the same grade, over 690.49. Both are labelled lows, not buy signals, and neither rate applies to any single name or sector fund: a downside band break on INTC, AMD or SMH has no measured bounce rate of its own.

Thursday, which had no edition of its own, is the other half of the story. The census rose (393 up, 255 down, median +0.71%, RSP +1.20%) while QQQ fell 1.42% and closed 715.34, under its daily floor, because the reports went into the cap-weighted names: GOOGL -12.64%, NOW -26.95%, FTNT -20.09%, MRVL -13.24% as the payer in the Lumentum report. Money rotated to XLV (+2.40%), LLY (+6.63%), XLU (+1.53%) and small caps on a bond rally after a 4.894% seven-year auction, and IWM closed over its weekly ceiling. That rotation is the month in one session. Friday took the small caps back (IWM -1.65%, back under 282.79), and the rest of the rotation held.

TAPE file gross $23.99B, about half structure · equity tape net +$0.8M: calls written over bought $199.6M, puts written over bought $200.5M · monetization neutral, 83.9% on a -2.3% trailing move · 81.8% of single-stock dark-pool volume at the closing auction, quoted, not counted

The October autopsy — what the month did, who sold, and what we saw too late

October was a rotation with a crash inside it, and the crash was in the AI trade, not in the market. SPY fell 0.91% while the equal-weight RSP rose 4.95%. QQQ lost 6.66%, the 33-stock AI basket 23.6%, SMH 14.21%. XLE rose 13.85%, XLP 10.94%, XLU 10.12%, XLV 9.77%, and nine of the eleven sector funds finished higher; only XLK (-8.30%) and XLC (-9.98%) fell. RSP beat QQQ by 11.61 points, the widest month in the 281 months since RSP listed in 2003. Across the 649 stocks and funds we can compare, 327 rose and 321 fell, median +0.20%. Anyone holding AI lived through a crash. Anyone holding the average stock had a good month.

The comparison that matters. We measured XLK against the average of the other long-history sector funds for every month since 1999. October’s gap is -16.43 points. Only seven months since 1999 show tech lagging by 10 points or more, and all seven sit inside the 2000 to 2002 bust. That is a comparison from one episode, not a rate: in the month after those seven, QQQ rose four times and fell three, with a range from -17.49% to +12.41%. It says the size of this rotation has one family of precedents, and that the month after it was violent in both directions.

The order on the bands. The index closing highs were October 6. XLU broke over its October ceiling on 10/9, XLP, XLE and XOP on 10/12. INTC and DRAM broke under their floors on 10/14, SMH, AMD, MU and XLK on 10/16, and QQQ only on the last close of the month. So the defensive breakouts came first, by days, and the index came last, by twelve sessions. Of the 41 lines on Silva’s October board, 17 finished under the floor and 10 over the ceiling. Every line that closed September over its ceiling (META, AMD, INTC, SMH) closed October under its floor, and six lines that closed September under the floor (XLY, XLF, XLU, XLP, XRT, TLT) closed October over the ceiling. October was September run backwards: in September RSP fell 5.18% while SMH rose 9.41%.

The late buyer

The buyer who was hurt worst was not the one who bought the July low. It was the one who bought the rally. From the closes of July 29, the Leopold low, when the Situational Awareness blowup cleared the crowded AI book, the 15 leaders below rallied between 12% and 87%. The median of the 15 has now given back 86% of that rally, and six are under their July 29 close. Only four ever closed at a new record after the low; for the other 11 the whole rally was a lower high. A buyer at the July 29 close is still ahead on 9 of the 15, led by MSFT +33.6% and MU +19.7%. The damage fell on whoever arrived late.

NameJuly 29 closeRally highOctober 30Rally given backBuyer on the first close over the old record
SMH504.22633.90522.4986%never reached, 5.2% short
MU739.001,097.39884.7459%never reached
NVDA190.01239.24204.6470%-14.3% from 238.90 on 10/5
AMD429.56649.42504.8066%-18.0% from 615.52 on 9/21
AVGO370.32427.76326.08177%never reached
TSM374.67485.80406.1372%-16.4% from 485.80 on 10/5
ARM224.89333.20205.97117%never reached
AMAT436.45548.15390.09142%never reached
ORCL117.74162.52130.6771%never reached
CRWV60.82107.7354.87113%never reached
NBIS148.22277.68154.1495%never reached
META585.61777.59614.9985%never reached, 1.6% short
MSFT390.54543.45521.9114%-4.0% from 543.45 on 10/13
GOOGL336.71377.65279.84239%never reached
INTC81.88127.3965.76135%never reached

Read plainly, the rally after the Leopold low handed the risk from the first buyers to the last ones. That is who held it when the month turned; it does not mean anyone timed it, as the section after next shows. The late call buyers can be marked directly: the GOOGL January 360 calls bought at 8.19 on 10/27 last traded 1.19; the MU January 900 calls bought at 139.47 on 10/28 (open interest confirmed the opening) last traded 82.00; the TSM December 440 calls bought at 22.50 on 10/28 last traded 8.35.

Was buying puts the only trade?

No, though on the AI names it came close. We measured the claim rather than argue it. The rule: a put about 5% out of the money with 30 to 45 days left, bought on the first session of each October week on 20 names, marked at Friday’s close. Entries come from actual prints on the options tape where one exists, otherwise a model price at that day’s implied vol. We checked 19 contracts against broker closes and the test runs about 5% rich.

SetPut, average multiple, winsSame-delta call, average multiple, wins
AI 12 (SMH, NVDA, AMD, MU, AVGO, MSFT, META, GOOGL, AMZN, ORCL, PLTR, CRWV)2.50x, 50 of 600.25x, 1 of 60
QQQ2.87x, 5 of 50.09x, 0 of 5
SPY1.18x, 4 of 50.18x, 0 of 5
IWM0.54x, 1 of 50.83x, 2 of 5
Shelter 5 (XLV, XLE, XLU, LLY, XOM)0.14x, 0 of 254.37x, 24 of 25

Four things temper the headline, and they are the answer to “the only thing we were supposed to do.”

Who sold, who bought

The selling at record pace is real. The timing does not show foreknowledge. The sellers with the clean timing were the companies themselves. The high was 10/5 on SMH and 10/6 on SPX and QQQ. Every bank figure below is as reported by the outlet named in Sources; we did not read the raw notes.

WindowWhat was reportedAgainst the high
JuneAlphabet (GOOGL) raised $84.75B of equity at about 355; margin debt peaked at a record $1.50Tfour months before
JulyGoldman’s prime desk saw $23B of tech selling, and margin debt fell by a recordat the Leopold low, SMH then rose 25.7%
Third quarterInsiders at AI and chip companies sold $8.3B and bought $21.6M, rising into Septemberbefore, on a standing condition
Week of the highJPMorgan saw semis de-risked; Citadel’s retail platform net bought $9.1B, $2.6B of it semisat
10/9 to 10/15Goldman: the fastest tech cut in its data from 2016, mainly longs sold; net leverage a one-year low of 48.6%during, SMH started 4.2% off its high and ended 11.8% off
Week to about 10/16BofA: a record $14.8B single-stock sale, tied in part to mutual-fund tax selling; tech funds took in $18.8B and SOXL a record $6.1Bduring
To 10/23All three Goldman strategy groups lost over 1% in a day on 10/22, a first since COVID; tech-focused funds about -10% on the month; two-week tech selling about $20Bduring
10/26 and 10/27Meta (META) raised its plan to $97.5B of equity, common at 625during, with more supply to come

Four facts stand against the idea that professionals sold because they knew. The same desk’s largest selling this summer marked the July low, and the market rallied for ten weeks after it. Tech-focused funds were down about 10% on the month, and funds with foreknowledge do not lose 10%. BofA’s hedge-fund clients bought for four straight weeks until the mid-October week. And BofA tied its institutional selling to tax-loss harvesting, and those institutions were buyers a week later.

What the record does support is a transfer. Stock moved from prime books to retail platforms and leveraged funds. Retail bought $9.1B in one week on Citadel’s platform, the semis slice at a 96th-percentile reading, a record week in the neoclouds, and a record single-name buy of CRWV, which has lost up to 10.6% from that close and sits 0.63 over the month’s closing low. SOXL’s share volume doubled while the fund finished the month -47.9%. Retail is not one actor: BofA’s retail clients sold in the same weeks Citadel’s bought. On the tape, after matched legs and stock-substitute trades are removed, call buying in the AI leaders ran $2.9B in the last week of October, level with the week of the high, with most of those names 15% to 25% lower. Put buying rose 64% only after the break.

The positioning is less crowded, not cleaned out. Net leverage of 47% to 48.6% is a one-year low, yet tech is still 23.5% of Goldman’s book, the 90th percentile of five years. The sellers with the clean timing were issuers and insiders: Alphabet sold stock at about 355 and GOOGL closed at 279.84; Meta priced at 625 and META closed at 614.99; insiders sold $8.3B in the quarter; buybacks are at their lowest share of market value since August 2021. That supply was public and it needed no secret. As for the valuations, the regime line carried the warning all month: the real 10-year over the breakeven (2.63% against 2.34%), a fair multiple near 11 times, the highest multiples first. The capital-spending story soured on funding, cash burn and guidance, not on demand: Alphabet, Amazon and Intel all reported AI demand up, and nothing on our disk shows a spending cut.

The signs, and what we did with them

Nothing separated this top in real time at the index. At the group level the price said so on 10/8, and we said so on 10/14. The record close on 10/6 came with a neutral gauge (50.4), credit improving on the day and SPX inside every band. What we can do is sort what was on the table.

Acted on. The list’s one index hedge, a December SPY put, went on at the 9/29 close. On Wednesday 10/14 the closes we had named printed (QQQ under 737.81, IWM under 273.83, SMH under 580.76) and we stated 19 bearish group calls: that is the date our reads turned. On 10/15 and 10/16 sold beats became a counted input. On 10/19 the first bearish AI structure went on our list.

Seen and set aside, with the reason we gave.

On disk, and read the other way. Rising spot and rising vol we read as extension, not a top, and that forward test failed on 10/5, one session before the record. Band-break bounce rates measured on the indices were cited on chip names they were never measured on, a correction made on 10/14. The AI-capex reporters’ own reactions turned two weeks before the spenders reported: from 10/14 to 10/23 one of ten closed higher on day one, mean -6.43%, while the tape-wide earnings label read mixed and then bullish.

Visible only afterward. Speculative themes peaked 9/8 to 9/25, ahead of the chips. Vol drained out of NVDA and MU into the high. CRWV’s 2032 notes widened about 100 basis points net of Treasuries while the stock rose to its peak, and we had no single-name credit read until 10/15. The prime and flow data above were published days to weeks after their windows.

What the lag cost. From the 10/8 close to the 10/14 close SMH fell 7.00%, the AI basket 11.97%, QQQ 2.49% and SPX 0.60%. By the time we stated the calls, 62% of SMH’s fall to Friday was done. Another 7.48% on SMH and 5.28% on QQQ came after. On the S&P 500 our no-direction read was right: the index lost 3.05% from its record and the money was not there. One more mark against us: the AI basket’s bearish call retired on a quiet bounce on Thursday 10/29, one session before Friday’s -8.16%.

Did the sellers tell us?

No, and that is the uncomfortable half of the answer. On September 5 we stopped reading sold calls, put buying and one-day dark pool at the bid as bearish. October was the first real test since: a 27.8% fall in the basket from its high. We ran each one against the basket’s next 5 and 10 sessions from 9/1 to 10/30, with 550 other stocks as a control, and with the 16:00 closing auction removed.

One month cannot settle this. It sits beside a 51-session replay of this year’s January-to-April correction and a 57-date backtest, and all three say the same thing. What worked is what we already call bearish: price breaking at the group level, a trend under the bands, members under their 10-session lows. Our job was to say so at 10/8. We said it at 10/14.

The analysts who read October correctly — the Maverick had the trade, Reminick the dates, Kramer the range

The right read in October was a rotation out of the AI build-out under a nearly flat index, not a falling market. Three desks had the three pieces, each on record before it happened: the Maverick had the trade, Allen Reminick had the dates, and Michael Kramer had the range. Being bearish was not what separated them. Tim Knight, the shelf’s most bearish voice, was right on AI and lost on his largest position.

We searched the 627 October commentaries on our shelf, read the files behind every call that mattered, and graded those calls against the closing prices. These are outside views, and none of them is a signal.

The month they are graded on

October was September run backwards. In September QQQ rose 3.21% and SMH 9.41% while the equal-weight fund RSP fell 5.18%. In October QQQ fell 6.66% and SMH 14.21% while RSP rose 4.95%. SPY lost only 0.91%. Our 33-stock AI basket fell a mean 23.61%, while XLE gained 13.85%, XLP 10.94%, XLU 10.12% and XLV 9.77%, and TLT rose 5.22%. RSP had no meaningful down stretch all month. So “buy puts” worked on the AI names and on QQQ. It did very little on the S&P 500, and the equal-weight S&P 500 rose.

The five desks that had it

1. The Maverick: the trade. He is a standing bear, and we use his framework and not his mood. But this call was more than his usual mood. His September 28 video was titled “October Outlook Buy The Losers, Short The Winners”, and on September 30 he named the list. On October 6 he said memory would crack in the order WDC, STX, SNDK, then MU, and the lines fell in that order between October 8 and 13. On October 8 he said SMH or DRAM was a better short than QQQ. His 19 listed names that we track rose a mean 6.69%. His method is a positioning read: when one group is crowded and the rest is washed out, the money turns. His misses were real too. Crude never settled where he said it had to, GOOGL and META broke before the end he drew, and we wrote on September 30 that his rotation failed its first session. It then ran for a month.

2. Allen Reminick: the dates. He uses fixed time cycles and calendar work. On September 29 he put the high at October 6 or 7. After the close on October 6, the day of the record, he said an intermediate top was days away. He then gave October 16 as a low, October 21 as a bounce high and late October 27 or early October 28 as the next high, and each landed within a session. His size was wrong: he said 10% to 12%, and the S&P 500 fell 3.05% close to close. He was bullish from August, so this was no standing bear turning right.

3. Michael Kramer: the range. On October 10 he said there was too little volatility and correlation to move the index, that it would stay between about 7,500 and 7,825 until the megacaps reported, and that sectors would move underneath it. From then to Friday the S&P 500 traded 7,527.56 to 7,821.67. No desk on the shelf drew the difference between the index and the stocks more exactly. He then gave the line for the rotation, SPY over RSP at 3.50, which broke on October 29.

4. Cem Karsan: the mechanism. He explained why the index stayed calm while single stocks moved violently: dealers compress volatility into a monthly expiry, and the support ends with it. He tied the AI build to debt, and pointed to 2000, when the Nasdaq went first while health care and energy held. His index call, higher into the vote, is behind so far.

5. Arete Trading: the turn. He was a chip bull who asked on October 7 why anyone tries to call tops. On October 8 he closed every semiconductor trade, and on October 12 his headline was to watch the S&P, not the Nasdaq. His gauge is the share of S&P stocks above their 50-day average divided by the same share for the Nasdaq 100. He also named forced selling from margin debt. Then he bought semiconductors on October 15, 21 and 27, and none of those held.

The next group. Darius Dale says his model turned risk-off in September, and he framed AI spending as a capital bubble, though he was wrong on bonds. Mike Silva marked the washout in the laggards on September 28 and 30 and kept the best daily level map on the shelf, but his regime label follows the tape by design. Rob’s Child read the top in yields from the chart on October 7 and 8, and then RSP and Treasuries, after whipsawing at the high. Knight’s island reversal on QQQ and XLK at the record was a real dated signal inside a permanent short book, and his largest short, XLU, rose 10.12%. Ilya Spivak had the bonds. Brent Kochuba had the volatility call, then went long tech on October 19. Benjamin Cowen’s midterm-year correction arrived, but his bond timing has not.

Who missed. James at InvestAnswers called the AI bubble fake and bought each leg. Click Capital said buy the dip at every stage, and on Friday said it again. Krown’s bitcoin target of 92,000 for October 31 met about 76,900.

What the indicators are worth

We rebuilt the three that mattered on our own data. The results are less flattering than the hit rate.

The rotation. Buying last month’s worst sectors against its best has no edge: 333 months, right 50.2% of the time, average gain zero. A narrower version matches what he saw. When technology beats the other sectors by 8 points or more in a month, the others won the next month in 13 of 18 cases, by a median 2.31 points. September’s gap was 10 points, so that test was live at the September 30 close. It is a modest lean, and it says rotate, not short. Among 592 stocks, September’s top tenth fell a mean 19.73% in October. The very worst tenth fell too, by 4.27%. The money went to the middle.

The dates. Reminick’s cycles failed our test. His 49-week cycle ranks 31st of 41 lags and matches market highs less often than chance does, 13% against 15%. Since August he has landed 8 of 10 dated turns within a session, but swings come so often on this tape that a coin does that about 5% of the time, and with a dozen desks publishing dates, one record like his is close to even odds. The October 6 date is harder to dismiss. We would use his dates as a clock and take size from the expected-move bands.

Breadth. This one was useful in real time. Arete’s ratio crossed its own 10-day average for good on October 8, two sessions after the record, and from that close SPY fell 2.35% and QQQ 7.64%. Inside our census, the AI basket’s share of names above their 50-day average went from 57.6% at the record to 3.0% by October 16, while the rest of the market rose from 31.2% to 45.6% by month-end. But it describes a rotation and does not sell the index. A fall this fast in SPY against RSP has happened five times before since 2003, and SPY was higher 20 sessions later in four of them.

Knight’s island reversal has appeared only twice near a record in QQQ since 1999. In June 2020 QQQ rose about 13% over the next month, so it is a dated observation with no rate. Margin debt explains why a fall feeds on itself, but it is monthly and late, so it cannot date one.

Luck against process

A desk that is bearish every month was right in this one by default, so the test is whether it changed its view on a dated signal. Reminick did, by a calendar set in advance. Arete did, within two sessions of the high. The Maverick’s call came with a long list and stated conditions, which is more than mood, though the crude condition failed and the trade worked anyway. Kramer’s range was derived, not felt. Knight and Wolfgang held the same view they always hold.

What they say now

None of the seven best desks says the low is in. Five of the seven tie their path to Tuesday’s vote.

Where they agree. Nobody on that list sees capitulation yet, and most of those who give a date beyond the vote point to early 2027 as the next large risk: Reminick has a hard January and February, and Karsan has a top at the January expiry.

Where they split. Reminick puts the low around November 3 and a rally into early January. The Maverick wants one climactic flush before he would buy the index. Dale says this is not the morning to buy, a convincing blue wave is the most underpriced risk he has seen, and the path looks like December 2018 before a sharp squeeze early next year. Karsan says stay hedged through the vote and cash in the hedges when volatility jumps. None of them has given a line on Friday’s payrolls.

What the tape supports. The rotation is still running: SPY over RSP closed at 3.46, its low of the month, and XLE, XLV, XLU, XLRE and XLB ended October at their highest closes of the month on a day 32 of the 33 AI names fell. No washout has printed: the FOM gauge sits at 50.0, VIX closed 19.42 after a 23.93 high, and we see no breadth thrust. Bonds are rallying with crude at 104.12, which sides with Reminick over Cowen on the long end. A post-vote rally is only a base rate: from election day to eight sessions later the S&P 500 rose in 12 of 19 midterm years, median +1.33%, and 2018 was the worst of them into year-end.

The levels. SPX 7,580.07 (zone 7,565.50 / 7,689.54 / 7,813.57, a reversed trend reading) has a daily band for Monday of 7,506.63 / 7,653.51, a weekly band of 7,413.82 / 7,746.32, a monthly band for November of 7,239.64 / 7,920.50 and a quarterly band of 7,089.48 / 8,213.60. Kramer’s double-top line, 7,507.77, sits just over the daily lower band, so one ordinary day can test it. Reminick’s 7,400 sits under the weekly band, so it needs a larger move than that week prices. QQQ 690.49 (zone 681.54 / 709.50 / 737.45) closed October under its monthly lower band of 702.09, a CONTINUATION on the INTC and AMZN reports, and its weekly band is 666.62 / 714.36. Our event card rules out new index-direction trades through Tuesday 11/03. We take no direction from any desk here, and none of this is a reason on its own to buy or sell anything.

The regime under it — a tightening Fed, a bond rally, credit soft by one print, and no washout

The regime did not change on Friday; the AI complex broke inside it. The Fed is tightening: it raised rates in September and held on 10/28 with two members wanting another hike. Yet the bond rallied into the month’s end: the 10-year closed 4.933%, down from its 5.307% closing high of 10/6, the 2-year 4.594%, the 30-year 5.309%, with Thursday’s 4.894% seven-year auction doing most of the work. So the fall in the chips came with falling yields, which is the signature of Silva’s 2018 comparison, bonds up and the Nasdaq down, two sessions running. Real yields remain the cap: the 10-year real yield 2.63% sits 29 basis points over the 2.34% breakeven and the fair multiple is about 11 times, so the market is doing the Fed’s tightening and the highest multiples go first. The Fed’s balance sheet shrank $8.0B on the week.

The dollar and metals. DXY closed 101.96, inside its zone, 45.3% of the 101.445 / 102.02 / 102.59 zone on the 10/29 sheet (range reading 69, trend 100.75 below price), 101.376 / 101.93 / 102.49 on the Monday sheet: no extension read. The strong-dollar block on commodities stays active, which stops new bullish metals positions and is no forecast on the dollar or on gold. GLD fell 3.47% to 355.78, under its daily 2-sigma floor and its weekly floor, with the dollar and rates flat. We have no named input for that move and we will not invent one; there is no measured bounce rate for it either. Crude settled near 104.12 on the front contract, over the Maverick’s 100 line, and XLE and XOP closed at October highs.

Credit. This is the point that keeps October from being a market-wide unwind so far. The public high-yield spread printed 3.26 for 10/29, 12 basis points wider on the day and 14 inside its 10/22 high of 3.40. HYG closed 77.43, higher on the week. None of the five lines we set on 10/22 for a liquidity drain printed on Friday: HYG held 76.39, USD/JPY held 155.27, three-month implied correlation sits at 15.60 under 20, and TLT rose while SPY and GLD fell. Stocks broke; credit did not confirm it.

Sentiment. The FOM gauge closed October at 50.0, neutral, down 8.6 on the day and up 5.4 on five sessions. It never left 38.3 to 58.6 in the month the AI trade broke, because it reads the average stock, and the average stock rose. No capitulation signal has printed, so nobody should read Friday as a washout.

Our event card. The inflation print on Thursday scored four of five inputs risk-on, and that card rules out new index-direction trades through Tuesday 11/03. Two of its inputs have aged (it scored VIX near 17 and SPX near its highs), but the card stands until its window ends, and the next top-rank print is payrolls on Friday 11/06 at 07:30 Central.

The count. On the AI groups three independent inputs now line up with the Fed: real yields over the breakeven, a bearish earnings regime inside the complex, and a price break through the October floors with the bottom test failed on both halves. That is convergence, and we state it plainly: bearish on the AI complex. On SPX and QQQ the count is one input with price at watch, so there is no index direction: the index sits inside every swing band and the equal weight sits over every moving average. Three bearish group calls stand (Gold & Silver Miners, retiring on GDX over 92.86; Semis — Equipment, retiring on SMH over 611.01; Gaming & Digital Platforms). The AI basket is one input short of a fourth, and it, Crypto and Copper sit in a cooldown to Wednesday 11/04. (Amended 10/31: five stand and no cooldown runs; see the note at the top.)

The four timeframes — the S&P 500 inside every swing band, QQQ through October with its quarterly floor now in range

The month’s bands closed, and the grade is cleaner than the headlines: the index never left its band, and QQQ left only on the last close. SPX finished 216.14 points over its October floor of 7,363.93 and never closed under it. QQQ finished at -1.31 monthly sigma, the Nasdaq 100 at -1.29. Beyond two typical moves on the month: INTC (-2.52), GOOGL (-2.11) and SLV (-2.08) on the downside, XLP (+2.82), XLV (+2.39), XLE (+2.15) and XLU (+2.13) on the upside. That is the rotation in one line: the two tails of the month were AI and shelter, and the index was the middle.

What the new bands do to next week. Silva’s new weekly range is 7,413.82 to 7,746.32 on SPX and 666.62 to 714.36 on QQQ. Neither the midterm election on Tuesday nor payrolls on Friday has a band of its own, so the width is the statement: 2.19% each way on SPX against 2.00% last week, a week that held the Fed and core PCE. Two lines matter, which is why we name them. QQQ’s quarterly floor of 669.25 now sits inside an ordinary weekly range, 21.24 points under Friday’s close: one typical down week would close QQQ through a band set for the whole quarter. And SPX sits at the bottom of its Monday zone (7,565.50 / 7,689.54 / 7,813.57, range reading -7, a reversed trend) with the trend value 7,676.15 over the whole of Monday’s daily band 7,506.63 to 7,653.51, so even a typical up Monday does not repair the trend. The first support test is the daily floor against the September 16 low at 7,507.77.

Moving averages say what the bands say. SPX closed under its 21-day (7,702.29), 50-day (7,697.08) and 100-day (7,614.99) averages and 3.7% over its 200-day (7,309.76). QQQ sits 1.6% over its 200-day 679.52. SMH sits 2.99 over its 200-day 519.50 and 18.27 over the Leopold low 504.22, with Silva’s weekly floor 492.57 under both. RSP is over every average it has.

The mega-caps and the roster — 190 names; INTC and RDDT the worst days on their records, the chips through their floors, and the shelter at its highs

The read this session: Friday punished guidance and funding, not demand, and the month’s split between the AI complex and everything else widened on the last day. We read 190 names this edition, each anchored to its own census close; a single-name downside break carries no measured bounce rate, so every break below is a continuation on a named event or a trend read, never a reversion call. Monday’s zones are from Silva’s new sheet.

INTC, RDDT, NOW and the report wreckage

**INTC closed 65.76, down 28.58%, its worst day since the 1970s and -45.30% on October, still about +78% on the year (zone 69.43 / 85.35 / 101.27, range -72).** It beat the quarter (EPS 0.52 against about 0.40, data-center and AI revenue +78%) and guided the fourth quarter to $15B to $16B against about $17.1B, on PC units squeezed by memory prices through at least the first half of 2027 and about $600M of revenue lost to the explosion at its Israel fab. The stated cause is the memory squeeze, so the AI build’s own side effect cut the old business. The Oct-30 92 straddle priced about 8.0% in Thursday’s last hour; the move was about 3.6 times that, the largest of its last eight reactions (measured median 7.9%). It closed under the zone low, and Silva’s new weekly band runs 60.06 to 71.46: a close back over 69.43 is the first repair, under 60.06 is a new leg. The tape wrote its long-dated puts by the 241,715 lots: willingness to own lower, never a direction.

**RDDT closed 91.73, down 29.22%, the largest down reaction on its record and the only name on our roster with all four bands new on one close (RDDT ZONE GAP); it sits inside the bearish Gaming & Digital Platforms group call.** It closed under its census quarter-end floor 96.44, and its census weekly band runs 81.51 to 101.95, so the first repair line is 96.44 and a close under 81.51 is a new leg. **NOW closed 105.12, -4.15% after -26.95% on Thursday, about 2.5 times the 10.6% priced, 33.9% lower on the week (zone 107.41 / 135.19 / 162.96, range -27.4, price under the zone). FTNT closed 136.08, -20.09% on Thursday (FTNT ZONE GAP). RBLX fell 18.05% on its report (RBLX ZONE GAP), SNDK 13.47% on a beat, a raised outlook and an 8-for-1 split (zone 1,119.12 / 1,301.04 / 1,482.96, range 2), and COIN 10.62% on its report and the Binance hack to 143.54 (COIN ZONE GAP).** Beats were not paid in the AI complex: that is the sentence the earnings section measures.

Alphabet, Amazon and Microsoft — the spenders

**GOOGL closed 279.84 and GOOG 278.22, -12.64% and -12.29% on Thursday and about -1.4% Friday, about twice what the options priced (GOOGL zone 275.83 / 306.54 / 337.25, range -10; GOOG zone 274.43 / 304.27 / 334.10, range -10).** GOOGL is under Silva’s expired weekly and October floors and -18.67% on the month. It sits over its Monday daily band’s midpoint and inside his new weekly 267.95 to 291.73, and GOOG is not carried on his boards, its census weekly row runs 264.64 to 291.80. A close back over 306.54 undoes half the gap. **AMZN closed 240.35, -1.41%, the tape’s one paid report: a low at 229.36, within a tenth of an expected move of Silva’s weekly floor, and a close inside every band, under a fifth of its 7.9% straddle (zone 237.46 / 250.04 / 262.62, range -2).** A close under the zone low 237.46 gives that repair back. **MSFT closed 521.91, -1.55%, four lower closes totalling 2.0%, inside every band and +1.76% on October against QQQ -6.66% (zone 516.15 / 528.10 / 540.05, range 12). It is the quality name that held, and Wednesday night’s +4.2% overnight mark turned into a -0.16% close: the reason we never grade an after-hours mark. META closed 614.99, -15.19% on the month, 1.6% under the 625 at which it priced new common stock on 10/27 (zone 565.92 / 637.34 / 708.75, range 24).** A close back over 645.04, Silva’s old October floor, is the repair line.

The chips — both halves of the bottom test failed

**SMH closed 522.49, -6.98%, 33.70 under its expired October floor 556.19 and -3.62 daily sigma, the largest miss on the roster (SMH ZONE GAP).** The bounce that ran from 10/23 to 10/28 failed its price half on Friday after failing its quality half for three sessions. Monday’s daily band is 506.52 to 538.46, the weekly 492.57 to 552.41, November’s 468.67 to 576.31, and the lines between are the 200-day 519.50 and the Leopold low 504.22. **NVDA closed 204.64, -4.22%, 5.49 under its expired October floor 210.13 and 0.95 over the 203.69 line of its vol-up-as-a-run-rolls-over tag (zone 202.22 / 213.65 / 225.08, range -1). It reports Tuesday 11/17 after the close. AMD closed 504.80, -10.88%, -17.48% on the month and under its 10/22 closing low (zone 481.94 / 542.62 / 603.29, range 14).** It reports Tuesday after the close, the Nov-06 straddle priced at 9.5% against a 6.7% measured median, reaching 456.97 and 552.63. **AVGO closed 326.08, -5.48%, its low through Silva’s October floor and its close back over it, the strongest large chip design on the month at -7.15% (zone 312.63 / 335.13 / 357.63, range 26). TSM closed 406.13, -5.95%, under its census October floor (zone 400.76 / 424.38 / 447.99, range 7.1). MU closed 884.74, -5.30%, leaving Silva’s October band on the last session (911.47) and -10.0% from Wednesday (zone 831.29 / 907.97 / 984.65, range 33). MRVL closed 196.40, -10.05% and -25.67% on the month, and LITE 840.70, -8.57% (MRVL ZONE GAP, LITE ZONE GAP); signed Lumentum terms, if any, and LITE’s report arrive this week.** ARM closed 205.97 (-28.89% on October, ARM ZONE GAP), AMAT 390.09 (zone 334.71 / 392.70 / 450.68), LRCX 239.11 (zone 200.97 / 241.77 / 282.57), KLAC 157.91 (KLAC ZONE GAP, its +5.45% report day given back on Friday), ASML 1,565.90 (ASML ZONE GAP) inside the Semis — Equipment call, and STX 644.58 (-30.11% on October, STX ZONE GAP), WDC 366.11 (WDC ZONE GAP), DRAM 48.01 (-20.46%, DRAM ZONE GAP) and SKHY (-7.95%, SKHY ZONE GAP) in memory and storage. QCOM fell 11.12% to 164.79, under its zone low (zone 166.22 / 176.77 / 187.32, range -49.6), before its Wednesday report.

The AI hosts, power and the speculative end

**CRWV closed 54.87, -9.66% and -37.02% on October, 0.63 over the month’s closing low (CRWV ZONE GAP); NBIS 154.14, -34.65% (NBIS ZONE GAP).** APLD, CIFR (-53.9% on October), BE, VRT, ETN, ANET (reports Tuesday after the close), COHR, GLW and ALAB all fell 4% to 11% on Friday, each without a measured bounce rate (their zones are gaps on Silva’s sheet except where named). ORCL closed 130.67, -6.37%, back at Tuesday’s 130.53: the investor-day break is void, though its 2031 bond held about 7.06%, so credit kept what the equity gave back (zone 120.27 / 132.27 / 144.26, range 42). PLTR closed 184.11, -6.27% with no event of its own, and reports Monday after the close, its front pricing 14.1% against a 10.0% median (zone 181.82 / 195.44 / 209.05, range -1.4). MSTR closed 130.51, -7.84% on its report, a CFO exit and the hack, under Silva’s expired weekly floor 132.52, over his expired October floor 128.27 on the close, with his new weekly 116.68 to 144.34 (MSTR ZONE GAP); IBIT 43.36, under its zone (zone 43.77 / 45.75 / 47.73, read from the sheet’s own row); HOOD 84.99 (HOOD ZONE GAP); bitcoin near 76,890, over the 73,975 line. SPCX closed 134.72, -6.37%, a hair under Silva’s October floor 134.80, and his new weekly runs 124.57 to 144.87 (SPCX ZONE GAP). TSLA closed 399.81, -0.87%, +12.68% on October, over Silva’s expired weekly and October ceilings on a merger bid, not on AI capex (zone 356.81 / 388.40 / 419.98, range 73); the merger’s math is where the commentary sits this week. NFLX closed 62.08, -10.78% on the month, inside his new weekly 59.53 to 64.63 (zone 58.50 / 64.50 / 70.50). AAPL closed 344.63, -1.23%, the one large mega-cap comfortably up on October at +3.49%, and reports Monday after the close, priced at 4.9% against a 1.9% median (zone 340.14 / 345.26 / 350.37, range 45).

The shelter — the other side of the month

**LLY closed 1,417.54, +3.19% after +6.63%, +22.51% on October, near its zone high (zone 1,231.22 / 1,332.00 / 1,432.78, range 115), and XOM 187.25, +2.16% on an EPS beat, +15.05% on the month and a 20-session closing high (zone 173.97 / 181.78 / 189.58, range 91).** LMT closed 636.95 (+25.08% on October, zone 570.40 / 626.33 / 682.25), LNG 330.32 (+22.78%, LNG ZONE GAP), CVX 235.52 (zone 220.02 / 230.33 / 240.63), ABBV 294.89 (zone 278.18 / 287.26 / 296.34, range 102), HII up 13.23% on its report (HII ZONE GAP), REGN up 13.33% on Friday, CMCSA, CME, ICE, DUK, D, NEE, SO, CNP, AEP, PEP, PG, COST, WMT, HD, LOW, MCD and V all finished at or near 20-session closing highs or ended the month up 8% to 19%. The risk to all of them is the same: XLP, XLV, XLE and XLU closed October more than two typical moves over their ceilings, and the Maverick, Silva and Karsan all flag the size of the move out of tech as the thing that snaps back on any AI rebound. A rotation that large is a trade to hold with a line, never one to add to here.

TAPE INTC Nov-20 70 calls x65,000 at the bid against Jan-15-2027 55 calls x41,000 (11:28, one roll) · NVDA Dec-18 190 calls x68,200 sold against Jan-15-2027 180 calls x66,000 (10:55, one roll, holders re-struck lower) · XOM December 190 calls x16,000 with a CVX roll in the same second · RDDT and INTC outside every band on the session

The sector map — nine of eleven funds up on the month, and the two that fell lost more than the nine gained

The month’s sector read in one sentence: XLK and XLC lost about $3.9T of market value at constant shares, the other nine sectors added about $1.8T, and the money stayed in the shelter through Friday’s close. We rebuild every sector from its members rather than trust the fund. On census equities 295 rose and 282 fell on October, median +0.26%. The crater, equal-weight: the AI basket -23.61% (cap-weighted -18.69%), neoclouds -33.10%, quantum -31.06%, crypto -27.60% against IBIT’s -8.41%. The shelter, equal-weight: energy groups +13.42% (28 of 29 higher), regulated utilities +11.49% (16 of 16), pharma +10.78% (14 of 14), staples +9.49%.

The rotation held through Friday, and it is at its widest. XLE over XLK ended October +22.15 points and XLV over XLK +18.06, both widest on the last session. XLE, XLV, XLU, XLB and XLRE closed at their October high closes, and XLP sat 0.63 under its 10/22 high. XLK beat XLV by more than 2 points on only two days of the month, 10/1 and 10/21. The one leg that failed is small caps: IWM -1.65% on Friday voids Thursday’s continuation over its weekly ceiling, though it still finished October +0.52%.

Three funds hide what their members did. XLB’s +8.30% sits on a materials median of -12.93%, a question of which members the fund weights, with copper and lithium names falling. XAR’s -1.68% hides defense primes at +7.93%. IGV’s -2.06% hides software names at a +3.41% median, with NOW, FTNT and FFIV the holes. XLF’s +5.82% hides banks at -1.67%. These are why we read members first.

The group calls had a bad two days and that belongs on the record. All 23 bearish group calls that retired this month retired on evidence fading, none on a price line, and 19 of them on 10/27 to 10/29. The ten that retired at Thursday’s close fell 6.35% on average on Friday. The AI basket’s call retired with SMH at 561.67, nowhere near the 611.01 line we had printed for it. A call that retires on two quiet closes inside a falling trend is not a repair, and our own language treated it as one. Three calls stand: Gold & Silver Miners, Semis — Equipment, Gaming & Digital Platforms. Crypto and Copper already meet the firing rule again and sit in a cooldown to Wednesday 11/04, and the AI basket is one input short with the same cooldown. (Amended 10/31: see the note at the top.)

Dark-pool labels missed the month entirely. The AI basket’s label net was +$16.70B on $288.9B of dark-pool volume, 81.1% of it at the closing auction, on a -23.6% month: labels lie and price does not. The usable dark-pool fact is magnitude: 61.3% of the AI basket’s volume came on down days against 33.6% for energy.

TAPE AI basket dark-pool label net +$16.70B on $288.9B, 81.1% at the auction · XLE over XLK +22.15 points, XLV over XLK +18.06 · ten group calls retired Thursday, -6.35% average on Friday · 3 bearish calls stand (5 as amended 10/31)

The options tape — the tape sold premium on both sides, and hedges moved lower and further out

On the day QQQ closed October under its monthly floor, the options tape did not panic. We strip structure before reading a side: of $23.99B of file premium, about 51% was packages and deep in-the-money stock substitutes, and the citable layer, $11.29B, was the largest of October’s 22 sessions. On that layer the single-name and fund tape came to +$0.8M: calls written over bought by $199.6M and puts written over bought by $200.5M. Take INTC out and it is -$146.3M. SPX was structure-dominated, non-directional, with its sided share 23.7% of gross and box-spread legs in December inside it; QQQ was two-sided at -$155.0M, -$6.4M without one line. The options net is pressure, never a direction, and Friday’s AI basket read makes the point: 32 of 33 names fell and its options residue was +$176.6M, because puts were written.

Near protection was sold and far protection was bought. By structure these are rolls, one vertical closed against another opened: index hedges were monetized and moved down, 283,259 lots of near puts sold against lower strikes with $60.7M of cash taken off, the most of October’s sessions, against a next-highest 186,028 on 10/16. The four earlier sessions over 130,000 lots were followed by QQQ -0.13%, -2.50%, +1.21% and +1.11%: no edge, and not a low call. The largest clean index put purchase came after the bounce: 30,750 QQQ June 2027 680 puts at 40.13 at 15:16 Central. The hedge moved lower and further out. It was not abandoned and it was not panic-bought.

Across the file, 83.9% of written-call premium was holders writing against stock they own, on a -2.3% trailing move: neutral, not the harvest of a run. 49 names read as harvest (TSLA, ORCL, LLY among them), none in AI memory, and MU moved out of it. The dashboards for 10/28, 10/29 and 10/30 were not captured, so there is no read of the dealer book, the 0DTE gamma map or the Flow Timeline for this edition. Our file-derived expiry ladder is a stand-in, not the panel: single-stock money rolled out to January on the NVDA and INTC call rolls, and QQQ’s November line went deeper negative. The hedge trade itself is crowded: SOXS and SQQQ carry the exhaustion tag, a rebound-risk description and nothing more.

TAPE stage-D gross $11.29B, the largest of 22 October sessions · calls written over bought $199.6M, puts $200.5M · index put roll-downs 283,259 lots, $60.7M off · QQQ Jun-2027 680 puts x30,750 at 40.13 (15:16) · monetization neutral 83.9%

Unusual trades — the structures worth naming

SMH — a 12,600-lot November put spread doubled in three sessions and reloaded 14% under the price

**A 12,600-lot November 525 / 480 put spread opened on 10/27 near 8.02 with SMH 550 to 557 and closed at 16.15 at 14:00:47 Central with SMH 514.23, about +$10.2M, into the session’s second leg lower.** By structure it is one vertical closed and two opened: December 450 / 400 x11,491 at 6.10 (11:58) and March 2027 450 / 400 x7,825 at 11.50 (13:09). SMH closed 522.49 (SMH ZONE GAP), Silva’s new weekly 492.57 to 552.41. That is the cleanest hedge monetization on the tape, rebuilt with more time and a lower floor: a desk taking profit on a hedge and staying hedged, never a buy signal.

TAPE SMH Nov-20 525 / 480 8.02 to 16.15, about +$10.2M · reloaded Dec-18 450 / 400 x11,491, Mar-2027 450 / 400 x7,825

QQQ — hedges rolled down into the low, and a June 2027 block after the bounce

**At 12:37 one ticket closed 14,775 Nov-13 725 / 690 put spreads at 22.66 (open interest matched lot for lot) and opened 26,595 Nov-13 685 / 650 put spreads at 8.66, about $10.5M of cash off the table, and at 15:16 30,750 June 2027 680 puts traded at 40.13, $123.4M, at the June money, 24.2 implied vol: the body, no wing premium.** QQQ closed 690.49 (zone 681.54 / 709.50 / 737.45, the sheet’s anchor is the after-hours mark). Put buying is hedging; these rolls are structure, and a Thursday trade rhymes with them: 35,000 January 650 puts opened against 25,000 November 700 and 690 puts sold, open interest confirmed the next day, up about 65% in a session.

TAPE QQQ Nov-13 725 / 690 x14,775 closed, 685 / 650 x26,595 opened (12:37, one ticket) · Jun-2027 680 puts x30,750 at 40.13 (15:16) · Thursday Jan-15-2027 650 puts x35,000 at 9.16

NVDA — a $370.2M one-second call roll, holders re-striking lower

**At 10:55:03 Central 68,200 December 190 calls were sold to the bid at 22.61 against 66,000 January 180 calls bought at 32.73, 43 implied vol.** By structure that is one roll, not new buying: holders stayed long and re-struck lower, and NVDA’s +$43.4M net on the day is that one second. At 13:40 70,000 November 230 / 260 call spreads closed against 35,000 November 215 / 237.5 opened. NVDA closed 204.64 (zone 202.22 / 213.65 / 225.08). Not bullish evidence: it says the holders are staying and paying to move down, which is what a trend does, not what a bottom does.

TAPE NVDA Dec-18 190 calls x68,200 sold, Jan-15-2027 180 calls x66,000 bought (10:55:03) · Nov-20 230 / 260 x70,000 closed, 215 / 237.5 x35,000 opened (13:40)

VIX — the November 19 calls sold at the high and re-struck to 22

**At 13:31 Central, with VIX near its 23.93 high, 41,659 November-18 19 calls went to the bid at 3.68 against 40,738 22 calls at 2.60, and from 13:44 to 14:38 21.5 puts traded at the ask (33,377, about 25,928 opening) against 21.5 calls at the bid: a synthetic short VIX.** By structure that is a hedge monetized at the spike and then a fade of it. VIX closed 19.42 (zone 15.27 / 18.20 / 21.13), and that is a tape betting the spike fails. Nothing here is a level for the index.

TAPE VIX Nov-18 19 calls x41,659 to the bid, 22 calls x40,738 (13:31) · 21.5 put and call pair x33,377 (13:44 to 14:38)

SPY — the largest month-end put prints are defined-outcome funds, not demand

From 15:00 to 15:08 Central the SPY October 2027 755.72 puts (Friday’s close as the strike), 642.36 puts (exactly 85% of it), 871.72 calls (a 15.35% cap) and 7.56 calls (1% of the price) traded as one reset, with sister lines in VOO, QQQ, IWM, EFA, LQD and HYG, $336.0M in the October 2027 series and $499.8M in January 2027. These are unsided, structure-dominated resets of defined-outcome funds: take them out before anyone reads month-end put premium as fear. SPY closed 755.72 (zone 754.16 / 766.45 / 778.74).

TAPE SPY Oct-29-2027 755.72 puts, 642.36 puts, 871.72 calls, 7.56 calls (15:00 to 15:08) · sister funds VOO, QQQ, IWM, EFA, LQD, HYG

INTC — the tape wrote the wreck

On a -28.58% day INTC traded $1.20B of premium; by structure, after its conversions, 241,715 long-dated puts went to the bid for $221.0M, at least 52,226 of them new writing, at the 5th percentile of its own long-dated vol history. At 11:28 Central 65,000 November 70 calls were sold at 2.07 against 41,000 January 55 calls bought at 13.00, a one-ticket roll down. INTC closed 65.76 (zone 69.43 / 85.35 / 101.27). Writing puts here is willingness to own lower, the tape’s opinion on price and never a direction: the post-gap study of 122 reports gives no timing edge either way.

TAPE INTC long-dated puts x241,715 to the bid, $221.0M · Nov-20 70 calls x65,000 sold, Jan-15-2027 55 calls x41,000 bought (11:28)

Volatility and mood — a hump over the vote, correlation climbing, and no fear gauge lit

The market priced an orderly AI repricing, not a crash. SPX at-the-money implied vol runs 16.3 for Nov-06, 14.1 for Nov-20, 15.4 for December and 15.8 for January: a hump over the midterm and payrolls, then a trough at November’s expiry. QQQ runs 27.9, 26.7 and 24.4, with its Nov-04 at 31.3, the vote. Silva’s VIX curve has November over December, 20.36 against 19.79. The SPX put wing prices 1.46 times the money on Nov-20 and 1.26 times on December. Fixed strikes on Nov-20 were flat over five sessions with only the put wing bid, and show a real bid over ten: hedging demand over two weeks, not a panic. Three-month implied correlation closed 15.60, up 2.75 in two sessions and at the 76th percentile, the one gauge that moved: the dispersion trade that kept the index calm while members moved apart is losing its bid.

41 of 65 measurable names sit at or above 1.10x front over back, against 17 of 42 on Thursday. 13 of those are reports, 4 the midterm, 3 payrolls, and 18 are post-print fronts still rich (HOOD 1.50x, RDDT 1.27x, INTC 1.25x, AMZN 1.21x, COIN 1.21x, IBIT 1.20x). Of the 12 large reports of the last two sessions, 8 closed outside their front’s 1-sigma, and all 8 were down. Next week’s fronts price a one-sigma report move of 4.6% on AAPL against a 1.92% median, PLTR 14.1% against 10.00%, AMD 9.9% against 6.73%, QCOM 10.4% against 5.73%, ARM 11.3% against 5.94% (from our measured reaction history). The FOM gauge sits at 50.0, 35.0 over the capitulation line: no contrarian input in either direction.

Is it too early or too late? From the gated evidence, in the market’s terms: late on the AI complex for anyone selling here (the floors broke, the hedge is crowded, and the rebound window opens Wednesday), early for anyone calling a low (no washout on any gauge, credit unconfirmed, call buying still level with the week of the high).

The commentary — the best read of October was on the shelf the day of the high, and it came from price, not from options prints

The commentary record for October is clear, and it is uncomfortable. The month’s best calls came early and were made from group price breaks, ratios and weekly closes; the desks still buying the dip had the worst two sessions. Forty commentaries dated Thursday and Friday were on the shelf, plus three that arrived late; every one was read, and what could be rebuilt was rebuilt on our own data. The index gave little: SPY finished October -0.91% and RSP +4.95%. The money was in AI single names and in the relative trade, RSP over QQQ by 11.61 points.

Mike Silva (FOM): October’s message was a rotation, and his label is back at Bear-Strong

Silva’s Friday deck returns his market-conditions label to Bear-Strong: SPX closed at 7,580.07, 104.93 under his 7,685 gamma flip, while the Nasdaq Composite, 25,437.00, sits under a falling 10-day average of 26,234.15. The words of his title, October Left an Important Message, are never spoken; his slides supply them: the first monthly bearish engulfing candle in QQQ since September 2021, semis -14.21 percent, staples over SPY through his warning line, and SPY only -0.91 percent. His ruling on the candle is that it is context, not a reason to go all short. Our event card rules out new index-direction trades through Tuesday 11/03, so his label is carried as a description of the tape, not as a trade.

The candle, rebuilt. His seven red bars reproduce exactly as a red outside month closing under the prior month’s low. The six earlier ones (June 2008, January 2010, October 2018, May 2019, February 2020, September 2021) left QQQ higher six months later in five and twelve months later in four, no worse than any other month, so his restraint is right. What the candle carries is a fatter left tail: a low 10 percent or more under its close inside six months in four of six, against 36.5 percent of all months. His negation rule is weak: September 2021 broke its high a month later and still fell 25.4 percent inside the year. The finding he left out sits in the same table. All six earlier months were market-wide falls, SPY down 3.63 to 8.81 percent and the equal-weight RSP down 3.31 to 10.16. October 2026 is the first with RSP up, +4.95 percent, and its 11.61-point lead over QQQ is the widest month since RSP began in 2003. In IWM’s record, QQQ down 5 percent or more with IWM higher had happened only in December 2000 and April 2002, both inside one bust: a comparison to hold, never a rate.

The month his deck does not show. On census closes 327 of 650 names rose, median +0.16 percent. Nine of eleven sector funds rose, led by XLE +13.85 and XLP +10.94; only XLK -8.30 and XLC -9.98 fell. Not one name rose in AI Infrastructure (median -26.90), Crypto (-25.20), Semis — Equipment (-24.84), Storage and Memory (-23.10) or Quantum (-31.51), while all 14 pharma names did. That is the fair answer to his title: the AI leaders broke under a flat index. Index puts paid little on that tape (SPY -0.91, IWM +0.52, RSP +4.95); AI puts paid heavily. Silva read where the money was going early, from his staples chart to CPB, PG and KDP. He did not read the top: his label stayed Bull-Strong through 10-13, with 8,000 named as upside on 10-09, and turned Bear-Strong on 10-14 with SMH already 10.71 percent off its 10-06 close.

The staples chart. XLP over SPY, 12-month rate of change, read 7.54, rebuilt to the hundredth, up from -8.68 a month earlier; XLP’s +10.94 percent is its best month since it began in 1998. Twelve earlier month-end crosses over his +5 line split by what the index was doing. With SPX already falling, 2000, 2002 and 2007 led to falls of 13 to 39 percent over twelve months; after long declines, 2011 and 2020 marked lows. This cross has SPX +5.15 percent over six months and 3.37 percent off its high, and its closest relatives (2013, January and July 2015) were followed by +19.3, -2.7 and +3.3 percent. A mild headwind; the warning is in the speed.

What only the deck shows. He says the day ended about half up and half down. That is the S&P 100 (44 up, 57 down, 74 percent of the volume in decliners). His own Nasdaq slide prints -2,355, the NYSE was 977 up and 1,717 down, the census 212 up and 437 down. His no-capitulation conclusion still clears his 90 percent bar. His VIX curve has November over December, 20.15 against 19.60 (settles 20.36 and 19.79, the widest gap in 12 sessions), a hump over the midterms and payrolls, with spot 19.42 under both and backwardation against the three-month VIX only at the intraday highs. Bullish percent is 57.80, mid-range; only its RSI, 77.57, is stretched. This deck has no FedWatch, positioning or dispersion slide.

The tick and the open. His first -1,000 NYSE tick came at 13:22 Central. The VIX high 23.93 followed, then SPY’s low 750.36 at 13:50 on a tick of only -569, and SPX closed 52.51 off its 7,527.56 low. His rule that an open outside the daily band is the usual road to a 2-sigma day holds on 52 sessions of his boards: 24 percent of such opens closed past 2-sigma against 4 percent of inside opens. SPX closed -2.79 sigma under Friday’s band: a weak REVERSION SETUP, graded by a Monday close over 7,580.07, measured 33 of 48 in this Fed regime. QQQ (zone 681.54 / 709.50 / 737.45) closed 11.60 under its October lower of 702.09: CONTINUATION, the input a company event, the INTC and AMZN reports.

His timers, rebuilt. The 5-10-20 stack, bearish since 10-16, is a volatility label: next-day moves of 1.34 percent against 0.72 in a bullish stack, with a small direction gap over 20 sessions. His bounce-from-extension idea (the Composite 1.70 ATRs under its 10-day) shows no edge at one to five sessions since 1985, though the last two touches, 03-27 and 07-29, paid well. The weekly MACD cross he calls pretty bearish has been followed by a higher SPX 26 weeks later 81 percent of the time. His year-end lean holds for the S&P 500 (higher in 78 percent of years since 1950) and not after a bad Nasdaq October: six NDX Octobers down 4 percent or more, two higher into year-end, median -5.98, 2018 among them.

His trades. There is no new list: the deck recaps staples longs he is trimming, and he warns a rotation back to tech could hit them hard. None is validated as a new entry. PG closed at 161.57 (zone 154.71 / 160.74 / 166.77): the 10-28 watch paid, and a fresh look is owed only on a pullback holding 160.74, void under the census iVol weekly lower 157.11. KDP closed at 35.84 (KDP ZONE GAP; census iVol weekly 34.22 / 37.46) and reports Tuesday before the open; nothing is owed before the print. CLX closed at 88.00 (CLX ZONE GAP; census iVol weekly 82.13 / 93.87) and reports Monday after the close. HD closed at 329.20 (zone 305.31 / 324.54 / 343.76): its watch holds over 324.54, void under 315.82. INTC’s -28.58 percent gap (zone 69.43 / 85.35 / 101.27) left a 131-session island, a single-name break with NO BASIS for a bounce rate.

What grades him. Monday his label stays Bear-Strong on an SPX close under the re-posted flip. Bull-Weak needs a close near 7,685, over Monday’s daily upper 7,653.51, and the flip sits with the trend 7,676.15 and the zone mid (zone 7,565.50 / 7,689.54 / 7,813.57), so one close over about 7,690 changes three reads. Bull-Strong also needs the 10-day to turn up: a Composite close over 26,008.86 Thursday or 25,955.34 Friday, or a gain of about 3.9 percent sooner. The first support test is Monday’s daily lower 7,506.63 against the 09-16 low 7,507.77. Frames: SPX week 7,413.82 / 7,746.32, November 7,239.64 / 7,920.50, quarter 7,089.48 / 8,213.60; QQQ week 666.62 / 714.36, November 647.08 / 733.90, quarter 669.25 / 810.29. AAPL reports Monday after the close, the midterms and AMD come Tuesday, payrolls Friday at 07:30 Central. QQQ over 762.86 negates the candle on his rule; the record says not to read that as an all-clear.

The Maverick: the call of the month

He is a standing bear; we use his framework and score it, and we do not adopt his mood. His Thursday recap, “Ugly Ugly Ugly,” named no option trade, so there is no print of his to check that night. His five morning ideas: META up on rotation out of GOOGL (+1.96%, right, given back Friday); a quick GOOGL bounce (a real morning scalp off the 280.36 low, near the lows by the close); fade LITE (+0.61% Thursday, -8.57% Friday); sell put spreads on software (mixed); gold loses 4,000 and the bull case ends (it settled 3,898.9 on Friday). His Friday map worked: QQQ support 695, target 690, close 690.49; SPY support 760, target 755, close 755.72. His 200-day call is unfinished: QQQ’s low 683.45 against 679.52.

The bigger record is 10-06, the day of the record closes. He said storage cracks first, then memory, then chip equipment, with the hyperscalers and NVDA last, and we fixed the lines that night. The first close through each line came in that order: WDC and STX 10-08, SNDK 10-12, MU, AMAT, LRCX and SMH 10-13, NVDA and SPY 10-14, META 10-19, AMZN 10-29; MSFT and AAPL never. Six of his seven dated calls hit (AAPL’s was the miss). From that close to Friday: INTC -41.5%, SNDK -30.7%, SMH -17.4%, QQQ -9.1%, while XLV rose 10.6% and TLT 5.9%. The caveat belongs beside it: he is right at every top, his July and August memory calls were early on MU and QQQ, and his RSP ceiling at 213 was wrong.

Who else had it, and who did not

Tim Knight called the QQQ island top the morning after it printed, then cut shorts mid-month, and his largest positions, short bonds and utilities, lost; he gave up the bond short on Friday. His records check: INTC’s -28.58% is its largest one-day fall on our data back to 1980, GOOGL’s -12.64% its largest ever. Rob’s Child read a rotation from technology into yield on 10-08 and went bullish RSP on 10-15; RSP is up 3.0% since the first call, QQQ down 7.6%. Benjamin Cowen’s 83,000 bitcoin line held up: two weekly closes under it, then the fall. Darius Dale said risk-off in September, and his “2% inflation is not compatible with winning the AI race” is the fiscal-dominance read we share; his 6.7% GDP excluding net exports checks by arithmetic. Cem Karsan called a window of non-strength after the 10-16 expiry and got it. Ilya Spivak called the record break a fake-out on 10-08, and his book scored five of six legs on Friday. Arete had the yield top but said the AI trade was back on 10-26, then turned on Thursday night. The wrong side: James (bitcoin 90,000 this month, the AI trade is back on 10-28), Mike Jones (the seasonal weakness ended 10-15), Geeks of Finance (QQQ back to 760). What the right desks shared: none of them got there from put buying or sold calls.

The 9% Nasdaq study, rebuilt

tastylive asked what follows a 9% Nasdaq drop inside a month. NDX reached exactly that on Friday’s close, -9.09% from the 10-06 high. Rebuilding his study from 1986, we get 102 episodes against his 100 and another 5% fall within a month 48% of the time against his 47%. His split, 68% in a crisis and 17% otherwise, also reproduces (72% and 16% on our own crisis list), but a crisis is known only afterward. Measured on what is known that day: with NDX over its 200-day and VIX under 25, as on Friday, it was 5 of 13, about 38%, higher three months later 7 of 13. That small group holds both the July Leopold low and March 2000. His own test, credit, is split: the CCC yield was near a three-year high at 17.83% on Thursday while the broad high-yield spread was a quiet 3.26.

Fridays, vol and the vote

Karsan says Fridays unpin vol. Since 1990 the session after a Friday down 1.5% or more does move more, 1.33% against 0.77%, but with VIX under 25 it rose 74% of the time over 61 cases. His own test, fixed-strike vol, did not fire: only the put wing was bid, and VIX closed 19.42 after 23.93. Brent Kochuba (SpotGamma) asked how long the rotation lasts if the Mags go, and Friday answered: RSP fell even as staples and energy held. His case for buying long-dated options is measured: one-year implied vol sits at the 7th percentile of its own history in GOOGL (34.5), the 12th in IBIT (43.3), the 15th in TSLA (45.6). Todd Horowitz called the sell-off orderly at midday; SPY’s volume finished at twice its 20-session average and VIX ran past his comparison, though SPX did close 52 points off its low. He expects a sharp bounce and would sell it, and he is a seller of bonds.

FX Evolution quotes 100% positive 12 months after a midterm: rebuilt, 19 of 19 since 1950. One month out it is 15 of 19. Volsignals warns the post-vote rally fizzles: the last three midterm Decembers fell, though December rose in 12 of 19 midterm years since 1950, and VIX eased in the week of 8 of the last 9 votes. FX Evolution’s 1989 analog is also real, the closest year for both the S&P 500 and the Nasdaq 100, but correlations of rising years are close to each other; it is a shape, not a forecast. Dale calls the midterm the risk the market has priced least, a blue sweep that slows data centers, and says December could look like 2018. For scale: SPX fell 19.8% from its record to 12-24-2018 and sits 3.05% off its record today.

Who pays for the build

Wolfgang’s doom leaderboard asks who drops first, and the month sorted the way he says. The five names burning cash (CRWV, GOOGL, META, ORCL, AMZN) averaged -15.8% in October; MSFT and AAPL averaged +2.6%. It is why the Maverick’s comparison to 2000 needs a qualifier: the generals did not fall together, they split by who pays. Then the suppliers followed: CAT -9.25% on a good report, the AI basket 32 of 33 down on Friday. His Friday and weekly figures for AAPL, AMZN and NVDA are wrong, but the sort is the sharpest on the page. Click capital makes the dip buyers’ case, forced selling on intact fundamentals, and bought INTC and likes RDDT; its Nasdaq facts check (the worst month since March 2025, -6.65%), but INTC closed -28.58% and RDDT sits in a group we call bearish. The forced selling itself fits the record: Dale’s cascading degross, the pod losses Kochuba cites and the prime-desk data elsewhere in this report all describe funds cutting while retail bought. Dividend Data and Dividend Talks made the same argument with valuation models; GOOGL went through the first’s 292 marker, and NVDA and STX closed under the second’s low cases a day after they were set. Options with Hans read the rotation well and memory badly. Trading Apologist alone asks how long, a year of grind for the hyperscalers on a time count, and his NVDA 200 level held at 200.92; his closing-hour accumulation signal was followed by Friday.

The one institutional trade on any desk’s page was a loser. tastylive’s $76M of calls checks: by structure it was single-leg buying, a run of 2,850-lot prints on Tuesday in the GOOGL January 360s, $38.69M on our tape, and open interest rose by about 39,000, so it opened. Those calls last traded 1.19 against an 8.19 average cost, about -85%. The AAPL January 380s ($35.03M) sit about -11% into Monday’s report.

The bond puzzle

Spivak’s point is the month’s most important regime fact on this shelf, and it rebuilds. From March to September crude and the 10-year rose together (correlation +0.50); in October -0.30. QQQ and yields ran -0.41, then +0.61. Stocks stopped falling on higher yields and began falling with lower ones, the signature of a growth or capital worry while crude keeps inflation alive. Knight reached the same place on Friday, asking why AMZN’s report lifted bonds, and answered that AI debt money is going home to Treasuries. Mike McGlone said hide in Treasuries and called metals stock puppets; gold’s correlation with QQQ rose to +0.60. Cowen expects the long end to bounce into the vote and fall after, the 2018 and 2022 shape; two sessions in, it fell first. Horowitz sees up to 6% on the 10-year by year-end; it closed 4.933%. Nanalyze’s 5.24% is three weeks stale. Karsan keeps gold for later on debasement, the fiscal-dominance read; on the day, bonds hedged and gold did not (-3.47%), the dollar gate on commodities is active with the dollar inside its zone, no extension read, and Arete, Spivak, Knight and McGlone are all against it near term.

Maps, and the hack

Arete’s Friday map was the best on the shelf: QQQ’s 700 lost, 693, then 685 to 688, with a 683.45 low; AMZN’s 230 with a 229.36 low; SNDK’s 1,150 against a 1,150.89 close. Its call that cybersecurity was the one constant lost (CRWD -4.10%). ET Tradytics had SPY 762 then 755 and got both. The Geeks' QQQ 700 failed; deep negative gamma brought more selling, not a reversal.

Cowen reads the Binance hack as narrative following price; bitcoin’s rebuilt 21-week average is 76,617, and Friday’s low traded just under it. Midterm-year lows came 71, 39 and 13 days after the vote, the same three dates Tyler S uses to say the low is in. Andrei Jikh reports a net loss near $1B and James calls it contained; COIN fell 10.62% anyway. James still adds to SPCX; his case is not independent confirmation of anything.

Where the desks leave Monday

Nearly everyone now agrees the AI trade is broken, which is a poor moment to add to that crowd, and no capitulation has printed: the FOM sentiment gauge read 50.0, neutral, through the whole break. For a session or two the measured odds lean up; for the month the rate regime and the CCC yield lean the other way. The event card, risk-on repricing, stands through Tuesday 11-03. Then ISM Monday, AAPL and PLTR after the close, the vote, AMD and payrolls. Knight and Wolfgang both expect Anthropic to pull its listing, set for the week of 11-16; that and NVDA’s report are the month’s real tests. One lesson from this week: after-hours marks were too kind on six of the seven names we tracked (GOOGL, NOW, MSFT, INTC, RDDT, AMD closed worse), and only AMZN closed better. None of this commentary is a trigger; every line above is a narrative weighed against price and the bands.

TAPE GOOGL Jan-15 360 calls 47,246 lots, $38.69M, single-leg, 10-27 Central · AAPL Jan-15 380 calls 71,627 lots, $35.03M, single-leg · LITE Jan-15 1050 puts 2,247 lots, $43.39M, 08:40 to 08:43 Central, a put position, not a direction · GOOGL deep-in-the-money puts 14:12 to 14:17 Central, $153.60M, stock substitutes, non-directional · INTC matched 70 packages in 2027, non-directional

Earnings — the AI reporters soured two weeks before the spenders printed, and beats stopped being paid

The tape-wide label reads mixed. Inside the AI complex it reads bearish, and that is the first number in this edition we wish we had printed on 10/15. Among the ten largest reports labelled on Friday’s close, six beats were paid (LLY, MA, MRK, LIN, TMUS, MO), three were muted (MSFT, CAT, KLAC) and GOOGL’s miss was sold. The split is complete: all four AI-linked reports closed under their pre-report closes and all six of the rest closed over theirs. Friday’s day-one reports read the same way, provisional until Monday’s close: INTC -28.58% on a beat, RDDT -29.22%, RBLX -18.05%, SNDK -13.47% on a beat and a raise, COIN -10.62%, against REGN +13.33%, RIVN +6.20%, MDLZ +3.85%, GILD +3.06%, XOM +2.16%, CVX +1.75%.

The market’s verdict on the AI spending story is the reaction, and its record by window is the month’s cleanest series. MU’s +3.03% on 10/1 was the last paid print of the old regime. From 10/14 to 10/23 ten AI-capex reporters reacted, one closed higher, mean -6.43% on day one: ASML and TSM on beats, then NOK, LRCX, BE, FIX, META, WDC and TXN. From 10/26 to 10/28 eight of eleven rose, mean +6.39%, led by VRT, SKHY, CLS and FLEX: the bounce inside the fall. On 10/29 and 10/30 three of twelve rose, mean -5.75%. The market’s question moved from demand to funding: nothing on disk shows a spending cut, suppliers beat, and the names sold were the ones paying for it or guiding down.

Reports against what was priced. 8 of the 12 large reports closed outside their front’s 1-sigma, all 8 down: GOOGL and GOOG (about twice the straddle), NOW (about 2.5 times), INTC (about 3.6 times the last-hour straddle), RDDT (about 2.4 times), MSTR, COIN and SNDK. MSFT, KLAC, LLY and AMZN stayed inside. AMZN is the opposite shape: a -4.61% open, a low -5.92%, a -1.41% close on a miss, under a fifth of its 8.06% straddle. Sizes are measured against our reaction history; the options priced a size and said nothing about direction, as they have all season. Our study of 122 post-print gaps found continuation 26% of the time, a fade 35% and chop 39%: no edge in timing what these gaps do next.

The week’s slate. AAPL and PLTR report Monday after the close; AMD, SMCI and ANET Tuesday after the close, with FISV, PFE, MPC, ETN, SHOP and KDP before it; NVO, MCD and UBER before Wednesday’s open, QCOM, ARM, COHR and LEU after it; LITE, TEAM, NET and AFRM on Thursday; OKLO and VST Friday morning. NVDA reports Tuesday 11/17 after the close. Left-tail lean, from our pre-print drawdown record (in-sample): names 20% or more under their 60-session high, 29.9% fell 8% or more on the print against 19.4% that rose 8% or more; names 10% to 20% under, 18.0% against 8.0%. PLTR, QCOM and NVDA sit in the second cohort. That is a tail skew, not a direction.

TAPE AI-capex day-one record: 1 of 10 higher, -6.43% (10/14 to 10/23); 8 of 11, +6.39% (10/26 to 10/28); 3 of 12, -5.75% (10/29 to 10/30) · 8 of 12 large reports outside their front’s 1-sigma, all down

Timing — the October chart landed six of its seven turns, and Monday decides the last one on its mirror image

Timing charts give timing and shape, never price targets, and both readings of the Savino October chart are graded on every dated turn. The original reading closed the month with six dated turns in a row, six of seven: a high on 10/6, a low on 10/8, the main low on 10/14, a high on 10/21, a low on 10/23 and a high on Wednesday 10/28, followed by the two lower closes it drew into month-end. The inverse reading has four hits and two misses. Its last turn, a low on Friday 10/30, needs a higher close on Monday to land, which would be one in a row and not citable; a lower Monday close is a miss on that turn. A turn lands within two sessions of a given date about 88% of the time by chance, so this is a clock that fit for a month, not proof of skill. The chart ends on Sunday 11/1, no November chart is on disk, and nothing is drawn past it.

The other clocks are counted once each and never as votes. The Maverick’s call for three up sessions and then a down leg late in the week landed (the Nasdaq 100 fell 1.38%, then 3.46%). Reminick’s end of the bounce on 10/27 or 10/28 landed, and his low around Tuesday 11/3 is pending. Helene Meisler expects the Nasdaq to turn overbought around the vote, the opposite sign, now from a lower start, and her line at QQQ 700 broke. Our own projection’s first low window, 10/22 to 10/29, was missed by at least one session: Friday’s close undercut the 10/23 low a session after the window shut. The Treasury-bond chart’s drawn high on 10/28 or 10/29 missed and its mirror’s low on Wednesday landed; the next drawn turn for both is Tuesday 11/3. Silva’s 2018 analog reached its eighteenth session with SPY 3.00% off its record, the closest yet to the 5% line that would engage its stock leg (740.14), two straight days of bonds up and the Nasdaq down, and credit not breaking; its next marker is the midterm. None of these sets a level: Monday’s band (7,506.63 to 7,653.51), this week’s (7,413.82 to 7,746.32), November’s (7,239.64 to 7,920.50) and the quarter’s (7,089.48 to 8,213.60) do.

Scorecard — grading Wednesday’s edition on Thursday’s and Friday’s closes (28.5 of 52 points)

Wednesday’s edition scores 28.5 of 52 points on Thursday’s and Friday’s closes, 55%. That is a C-plus. There was no Thursday edition, so it stood for both sessions. The arithmetic says the bottom of a B-minus; we hold it a step lower because two of the frames in its bottom line failed inside those two sessions. Counting a partial as a miss it scores 43%; counting it as a hit, 67%; without the option counts, 43%. The run of grades: B for 10/21, then B-minus for 10/22, 10/23, 10/26 and 10/27, and now C-plus for 10/28.

The Top Trades record. The list made about $1,636 on Thursday and Friday at the broker’s closing marks and stands at about +$384 on sixteen entries since 9/28, its first positive reading since the running record began on 10/8: twelve closed at about -$871, four open at about +$1,255, against about -$1,252 on Wednesday. Two lines printed on Thursday’s close and both came off by their rules: the AMZN 260 calendar retired on a close under 244.90 (243.79) at about 5.00, -$124, before its report, and the XLU 40 / 43 spread came off at its 43.00 take-off (43.14) at about 2.14, +$69. Still open: the SMH January 550 / 500 put spread 23.93 against 16.00 (+$793), the SPY December 750 put 13.93 against 11.52 (+$241), the SMH 500 put calendar 16.73 against 14.74 (+$199), the NVO 40 / 45 spread 1.70 against 1.48 (+$22). The INTC calendar lapsed unentered on Thursday, which kept the list off a -28.58% report, about $580 at the marks. The QQQ January 700 / 660 put spread did not enter: its level printed on Friday, two sessions before its date lock opens, about $564 not taken, and at 13.85 it has to be re-priced, never chased at a band-break low. The gain is one tape and mostly marks on put structures. Grade for the two sessions: A-minus.

The October report card — a B-minus that did not see the month’s one real miss

October’s 21 editions averaged a B-minus, 57%: one B-plus, two Bs, thirteen B-minuses, three C-pluses and two Cs. The best was the 10/14 edition (B-plus, 71%), the hot-inflation night and our first bearish call on the AI groups; the worst the 10/16 edition (C, 41%), the expiry night. The Top Trades list went from about -$94 on 10/9 to -$1,048 on 10/14, -$1,252 on 10/28 and about +$384 at Friday’s marks.

The leans, counted from the bottom lines, show where the read did and did not move, which is the honest measure of the month. On the index: no direction in 21 of 21, none bullish, none bearish. On the AI complex: two bullish (9/30, 10/2), eight neutral (10/1, then 10/5 to 10/13), eleven bearish (10/14 to 10/28).

Did we get October right? On the index, yes for SPY: it finished -0.91%, never closed outside its October band, RSP rose 4.95% and the median stock rose. QQQ’s -6.66% we left uncalled. On the rotation, yes, from 10/8 (“The Leaders Fell and the Average Stock Did Not”). On the AI complex, late. SMH’s highest close was 633.90 on 10/5. On 10/6, at SPY’s and QQQ’s high closes, we wrote “It is not a top signal.” The first warning came on 10/8, then three sessions of bounce labels that failed and a 10/13 line that the losses had moved to the banks. Our first bearish bottom line came on 10/14, seven sessions after the high close, with SMH 10.91% off it and 62% of its October fall already done; ten of the eleven structures on our list still leaned long, and five retired that night for about $1,111. From that call to Friday SMH fell a further 7.48% and QQQ 5.28%.

What the grades hid. The ten editions neutral or bullish on the AI complex averaged the same B-minus as the eleven bearish ones (2.62 against 2.65 on a four-point scale). The grade scores next-session lines, so it never marked the month’s one real miss, and the option counts carried it: without them our analytical lines scored 42% across the fourteen editions that state the split. The number October deserves to be judged on is the 62%.

Was this simply a month to own puts? On the AI complex, yes: the four put structures on our list made about $1,552 while the other twelve entries lost about $1,168. On the index, no: our SPY put sat $263 behind on Wednesday and made its whole gain in the last two sessions, and the month’s largest fund gains went the other way.

Monday and the week — the calendar, day by day, with the lines each session can test

The week holds a top-rank print, the vote and the last big reports, and our event card rules out new index-direction trades only through Tuesday. Silva’s Monday anchors post with his 08:15 Central brief; everything below is his bands, his zone sheet and the census.

The forward battlegrounds, each a binary with a trigger. The rotation: holds while RSP stays over its 21-day 215.65; fails on a close under it with SMH rising. The AI complex: a rebound into the zones in the window is the entry the map waits for; a weekly close under 503.63 on SMH breaks the Leopold low and voids the wait. The index: inside every swing band until a close under 7,413.82. Credit: an HYG close under 76.70 with the spread over 3.12, or two liquidity-drain lines, changes the regime. A washout, the one thing that has not printed: gauge under 15, VIX over 35, or a visible forced seller.

Bottom line

October broke the AI trade and left the market standing. RSP rose 4.95%, nine of eleven sector funds rose, SPY lost 0.91%, and QQQ lost 6.66% with the AI basket down 23.6%. The size of that gap has no precedent outside 2000 to 2002. Bearish on the AI complex, stated plainly: three inputs line up with a tightening Fed, the floors broke on company events, and the bottom test failed both halves. No index direction: SPX sits inside every swing band, credit has not confirmed the break, and our event card rules out new index-direction trades through Tuesday.

The story that it was all puts is half right; the story that Wall Street knew is not on the record. AI puts paid 2.50x and QQQ puts 2.87x, shelter calls paid 4.37x, and three-quarters of the AI puts’ gain came in the last two sessions, after a rule that lost in July, August and September. The hedge funds sold at the fastest pace in Goldman’s data, during the fall and at a loss. The clean sellers were the issuers: Alphabet at 355, Meta at 625, insiders at $8.3B against $21.6M bought. The late buyer paid: the median leader gave back 86% of its rally from the Leopold low. Our own miss is measured too: the price said so on 10/8 and our bottom line said so on 10/14, with 62% of SMH’s fall done.

What would change the read. On the index, a close under 7,413.82, or a credit break with HYG under 76.70. On the AI complex, a close over SMH 611.01. On the low, the one thing that has not printed: a washout, gauge under 15, VIX over 35 or a visible forced seller. Until then a rebound into the zones from Wednesday is the entry the map waits for, and chasing Friday’s low is not.

Top Trades to Follow

The institutional-flow list the next edition grades — defined risk, the body not the wing, owned for a window and retired at a line. Each entry states where the name’s long-dated vol sits in its own history, a level with its number, and an exit or retire line with its number. Our entry check is an entry-quality test, not a winner-picker: its pre-registered test showed no edge as a picker. Marks are the broker’s Friday closing marks, the market closed. The list stands at about +$384 on sixteen entries since 9/28. 41 of 65 measurable names sit over the front-rich line, so the list leans on calendars: sell the front the tape inflated, own the back body. No index structure is added on direction while our event card runs (through Tuesday 11/03), and no short premium is held into a report.

SPREAD SMH Jan-15-2027 550 / 500 put spread — CARRIED at about 23.93, +$793 from its 16.00 entry. The body at 550 financed by the dearer 500 wing, about 1.06 times the body at entry. Long-dated SMH vol sits in the middle third of its own history (44th percentile): a spread, which this is. SMH closed 522.49, 33.70 under its old October floor 556.19 and 2.99 over its 200-day 519.50, SMH ZONE GAP. Owned into the rebound window and into January; taken off on a weekly close under 503.63, retired on an SMH close over 611.01, the group call’s line, or at the Friday 2027-01-08 close. Maximum loss as held: about $2,393, its mark.
CALENDAR SMH Mar-19-2027 / Dec-18 500 put calendar — CARRIED at about 16.73, +$199 from its 14.74 entry. We own the March 500 put and sell the December 500 put, the front about 1.05 times the back like for like. Long-dated SMH vol sits in the middle third of its own history (44th percentile). SMH closed 22.49 over the strike, SMH ZONE GAP, Silva’s weekly floor 492.57. Owned through the rebound window into the second leg after the December expiry; the short leg is bought back on a weekly close under 503.63 or by the Friday 12-11 close; retired on an SMH close over 611.01, void on a weekly close over 639.47. Maximum loss as held: about $1,673.
PUT SPY Dec-18 750 put (the index hedge) — CARRIED at about 13.93, +$241 from its 11.52 entry, nothing added. Its 15.4-area December money and a 1.26 times put wing keep it a near-the-money body. Long-dated SPY vol sits in the middle third of its own history (61st percentile): the held put stays, no new single. SPY closed 755.72, 1.56 over its zone low (zone 754.16 / 766.45 / 778.74). Owned into the vote and payrolls; the SPY Nov-13 750 put is sold against it on the first SPY close over its daily upper (Monday’s 763.40), never at a band-break low; retired on a SPY close over the weekly upper 772.27. Maximum loss as held: about $1,393, its mark.
SPREAD NVO Jan-15-2027 40 / 45 call spread — CARRIED at about 1.70, +$22 from its 1.48 entry. The legs price about 1.0 times the January money. Long-dated NVO vol sits in the low third of its own history (0th percentile). NVO closed 40.00, at the long strike, inside its census weekly band 36.34 / 43.66, NVO ZONE GAP; it reports before Wednesday’s open. Owned into January 2027; retired on a monthly close under 34.15 and closed before its last 60 days, by Monday 11-16. Maximum loss as held: about $171.
CALENDAR XLE Jan-15-2027 / Nov-20 70 call calendar (new) — the rotation’s long side without chasing it: long the January 70 call at 3.53 (25.3 implied), short the Nov-20 70 call at 2.12 (30.9 implied), net 1.41, the front 1.10 times the back on the tape. Long-dated XLE vol sits in the high third of its own history (69th percentile): a spread, never a single, and the run bars naked long premium, not a calendar. XLE closed 70.02 (zone 66.74 / 68.75 / 70.76, range 86), 2.54 over Silva’s weekly floor 67.48. Follows the XOM December 190 call block. Owned into the Nov-20 expiry; retired on an XLE close under 67.48 or at the Friday 2027-01-08 close. Maximum loss: $141, the debit.
CALENDAR NVDA Jan-15-2027 / Nov-20 205 put calendar (new) — long the January 205 put at 14.30 (40.3 implied), short the Nov-20 205 put at 9.35 (48.9 implied), net 4.95, the front 1.21 times the back: sell the dated front at the body, own the back, no wing and no call sold. It holds both legs of the map: the sold put gains on a rebound into the window, the January put stays for the leg after it. Long-dated NVDA vol sits in the low third of its own history (23rd percentile). NVDA closed 204.64, 0.95 over its 203.69 tag line (zone 202.22 / 213.65 / 225.08, range -1). Owned into its Tuesday 11-17 report: the short leg is bought back by the 11-17 close, or lifted when NVDA trades 225.75 inside the window; retired on a weekly close over 246.63 or under 190. Maximum loss: $495, the debit.
WATCH AMZN Jan-15-2027 / Nov-06 240 put calendar (WATCH) — WATCH at about 8.15 (long 12.85 at 32.1 implied, short 4.70 at 38.9, 1.21 times); AMZN has reported, so the front carries the vote and payrolls, no print. Long-dated AMZN vol sits in the low third of its own history (29th percentile). AMZN closed 240.35, 1.2% over its zone low (zone 237.46 / 250.04 / 262.62, range -2). The missing point is the level: entered only on a close at or under 237.46. Owned into the Nov-06 expiry; retired on a close under 229.70 or over 251.00. Maximum loss: $815, the debit.
WATCH QQQ Jan-15-2027 700 / 660 put spread (WATCH, re-keyed) — WATCH at about 13.86, against 8.22 on Wednesday: the body at the money’s vol (23.4 implied), the dearer wing sold (26.5 implied, 1.13 times). Long-dated QQQ vol sits in the middle third of its own history (55th percentile): a spread, never a single. Its October-floor trigger printed before its date, and at Friday’s price it would be chasing a band-break low, so it is re-keyed: entered only on a rebound into 709.50 to 714.36 (the zone middle to Silva’s weekly ceiling) from Wednesday 11-04, after the event card ends (zone 681.54 / 709.50 / 737.45). Owned through November and December into January; retired on a QQQ close over 733.90, Silva’s November ceiling, or at the Friday 2027-01-08 close. Maximum loss: about $1,386 at the mark.

Left off on purpose: any new index structure inside the event card’s window; the rebound map’s other put spreads, every one still under its first zone; calendars on AAPL, PLTR, AMD, SMCI, QCOM and ARM, whose fronts are reports (INTC and RDDT are why no short premium is held into a print); longs in AMD, MU, NVDA, IBIT, NBIS and WDC, which pass the entry check but sit under a bearish count; and metals, under the strong-dollar block.

Sources

Expected moves: Silva’s daily expected-move boards for 10/29, 10/30 and 11/02 with the TradingView data files, his zone sets for 10/29 and 11/02, the ZONE DOCUMENT and the Zone Visual for each (no 10/30 zone sheet was posted), his weekly boards for 10/26 to 10/30 and 11/02 to 11/06, his October and November monthly boards, his fourth-quarter board with the year-end collar levels, the FOM sentiment gauge for 10/30 (no 10/29 gauge was placed; the 58.6 is derived from the 10/30 table) and our sentiment tracker, and our option-implied band ledger computed from Friday’s options file. Tradytics: the 10/29 and 10/30 options and dark-pool files, every October options tape for the put test, and our flow, dark-pool, monetization, setup and long-dated vol ledgers; the options dashboards for 10/28, 10/29 and 10/30 were not captured, so no panel is read in this edition. Recon census: all 651 per-ticker files for 10/30, the census of every October session, the sector chunks and summary, the name-behavior ledger and its roster, the census screen, the AI and speculative basket digests and their histories, the hourly cyber and software top watch, and the group-convergence screen and its call log. Intraday: the 10/29 and 10/30 close recaps, the pre-open and four intraday briefs of both days, the level frames, the scheduled-prints files, the news and sector sweep files, the FOM and Maverick morning briefs. Timing: the Savino October charts and their inverses, the Treasury-bond charts, the Maverick’s four-stage chart and our timing ledger. Commentary: Mike Silva’s 43-page deck of 10/30 (every page read as an image) and his transcript “October Left an Important Message”, the 40 polished transcripts dated 10/29 and 10/30, and the 627 October-dated transcripts on our shelf for the analyst reconciliation, plus our earlier syntheses of them. Earnings: our measured reaction history, the post-print continuation study, the pre-print drawdown record and the broker’s earnings calendar.

Web: Bloomberg, “Hedge Funds Cut Tech at Fastest Pace in a Decade, Goldman Says” (10/19), and “Retail Traders Boost Semiconductor Stock Buying Before AI-Driven Selloff” (10/22); Reuters via Investing.com, “Global hedge funds hammered by AI selloff this week” (10/16) and “Hedge fund returns worsen this week as tech selloff deepens” (10/23); Reuters, “Trend hedge funds saw worst week since 2024” (10/26); Investing.com, “Hedge funds cut tech stocks at record pace, Goldman says” (10/19), and Michael Lebowitz, “Hedge Funds Sell, Retail Buys” (10/27); Business Insider, “Retail Traders Are Piling Into AI Stocks” (10/16); 24/7 Wall St on BofA’s client flows (10/23); Forbes, “Hedge Funds Sold $20 Billion Of Tech In Two Weeks” (10/26); MarketWatch via Morningstar and etf.com on SOXL flows; CNBC and Advisor Perspectives on margin debt; Reuters and JPMorgan on Alphabet’s equity raise, Meta’s investor releases, Reuters and CNBC on Meta’s plan; Form 4 compilations of insider sales; Intel’s third-quarter release; CNBC on the Intel and Amazon reports (10/29 and 10/30); RTE on Intel (10/30). The bank figures are as reported by those outlets; we did not read the raw notes.

Connectors: the broker’s market-data quotes (option closing marks for the carried and candidate structures, the market closed), TradingView daily bars (rates, the dollar, crude, VIX), Yahoo daily closes (moving averages, the fund history behind the sector-gap comparison) and FRED (the high-yield spread, real yields, breakevens and the Fed balance sheet). Prior cycle: the 10/28 comprehensive analysis, daily report and regime snapshot.