The week ahead — October 18, 2026: Duration rallied and stocks fell anyway

Wed 10/21: Tesla and Lam Research after the close. Thu 10/22: Intel. Wed 10/28: FOMC decision, no projections; CME priced about 52% for a hike after CPI on October 14. Thu 10/29: September PCE. October payrolls and CPI land in early November.

The house view

Ahead this week, Tesla, Lam Research and Intel report inside three sessions, and they decide whether Friday's chip break stays a positioning event or becomes a financing one. The index enters under its 50-day with CME December hike odds at 74.3% and Brent at 109.86.

• SPY closed Friday at 763.25, down 1.97% on the week and under its 50-day; QQQ lost 5.52% (Coil bars through October 16).
• September CPI printed 3.7% against a 3.6% poll, with core 0.3% on the month against 0.2%; CME October hike odds moved to about 52% from about 22% (Reuters via Kitco, October 14).
• CME December hike odds are 74.3%, up from 50.2% a month earlier, and Brent settled at 109.86, up 3.7% (stockti citing AP, October 16).
• The ten-year eased to 5.09% and TLT gained 2.65% on the week while the Nasdaq-100 proxy lost 5.52% (AP; Coil bars).
• BofA's October survey had a net 37% of managers overweight tech, the most since August 2020, with cash at 3.9% (survey coverage, October 13).

Week horizon: 4 right, 1 wrong, 2 mixed over 7 graded notes. Month, quarter and year: 0 graded, 21 too\_early across those same 7 notes. The archive holds 7 prior notes, dated 2026-08-30, 2026-09-06, 2026-09-13, 2026-09-20, 2026-09-27, 2026-10-04 and 2026-10-11; the first had nothing to grade.

The setup

Transition. SPY enters the week under its 10, 20 and 50-day averages, with the 50-day at 768.50 the first line to reclaim and the 200-day 5.3% lower at 725.01. A Friday close back over the 50-day says the break was a shakeout inside a range. A close under 748 says the range itself has changed. QQQ's daily stochastic reads 4.6 and IWM's 3.8, so the tape arrives oversold with no reclaim yet, and the small-cap ETF is the one leg already below its 200-day.

What is actually going on

Into the week the market carries an open question: why bonds and stocks are moving apart, and which scheduled earnings decide the explanation that survives. The ten-year Treasury eased to 5.09% on Friday from near 5.25% a week earlier, per AP, and TLT gained 2.65% on the week. The Nasdaq-100 proxy lost 5.5% over the same days, and the Nasdaq composite had its worst week since June, per stockti. Last Sunday's note said the long end had stopped moving the index. This week showed where the pressure went. It moved from the price of money to the supply of it.

The supply is the AI build. Hyperscaler debt issuance has reached 159 billion for 2026, about 91% of the full-year forecast, and AI-linked issuers pay roughly 115 basis points over Treasuries against 78 for broad investment grade, per aiweekly citing Goldman. A fund manager survey from BofA on October 13 found a net 37% of managers overweight technology, the most since August 2020, with cash at 3.9%. The same survey listed a disorderly rise in bond yields as the top tail risk and an AI bubble bursting as the next. The most crowded position met the risk it had been surveyed on. Micron lost 5.6% Friday, AMD 6%, Intel 5.8% and Broadcom 4.5%, while the Dow fell 0.3%. The damage was concentrated, and the volatility proxy is still far below its 200-day, so the selling has been orderly.

The policy side did its part. September CPI printed a 3.7% headline against a 3.6% poll, with core at 0.3% on the month against 0.2% and core services at their largest gain since April, per Reuters carried by Kitco. Gasoline rose 9.1% in the month and supplied more than half of the headline move, which Reuters ties to the war with Iran. October hike odds jumped to about 52% from about 22% on CME's tool, and December odds sit at 74.3% against 50.2% a month earlier, per AP-derived stockti. The committee raised to 3.75% to 4.00% in September, and Goldman and Citi both now put the next hike in December, with BofA at 50 basis points more by year end, per the Reuters factbox.

Oil is the part the committee cannot answer. Brent settled up 3.7% at 109.86 on Friday, per stockti citing AP, against 102.12 on October 8, per straits.live. That brief also had four transits through the Strait of Hormuz on October 4 against a pre-crisis baseline of 85 a day, with the IRGC saying the strait stays shut until its conditions are met and CentCom saying the opposite. A supply shock is not something a rate rise can fix, and the September minutes already named oil, tariffs and the AI build as three sources of demand pressure, per Benzinga.

The live argument among strategists is whether the chip break is a valuation reset or a financing problem. The bulls point to third-quarter index earnings growth above 30%, per LSEG, and to a bank group whose trading desks posted records. The bears point to the bond market. A print of 6.44 billion in equities revenue at Goldman, up 72%, did not stop Morgan Stanley from losing about 3% on the day it reported 5.8 billion, per HeyGoTrade and the Morgan Stanley 8-K. Good numbers that sell are the same tell the note flagged last week. The market is paying for what comes next and not for what printed.

The market also keeps failing to broaden. The small-cap ETF closed a sixth straight weekly decline and sits under its 200-day. If the old story had been rate relief, last week was its week. It was not, which leaves oil and financing as the two live explanations, and both are slower to resolve than a data print.

Where this sits against consensus

On the policy path we sit with the street. Goldman and Citi put the next hike in December, and BofA has 50 basis points more by year end (Reuters factbox, mid-October), which matches CME's 74.3% for December (stockti citing AP, October 16). Where we part from the consensus read is the transmission. The standing assumption was that a lower long end would rescue growth stocks, and last week it did not: the ten-year eased and QQQ fell 5.5%. BofA's survey on October 13 had the heaviest tech overweight since August 2020, so the street was positioned for the rescue. On the index, Yardeni's 7,900 from September 16 and Citi's 8,100 are both further away than they were. We would be early if third-quarter guidance restores the earnings case quickly, and late if AI-linked spreads keep widening.

Scenarios

Central — 45%

Earnings hold guidance, oil stops rising, and SPY chops between the 20-day low and the 10-day while small caps stay heavy and the chip names stabilise without a clean reclaim.

Trigger. Tesla and Lam Research hold guidance on Wednesday 2026-10-21 and Brent closes the week inside 100 to 112.

Invalidated by. A Friday 2026-10-23 SPY close outside 748.00 to 776.00.

Upside — 20%

The chip break reads as positioning, Intel and Lam hold up, and SPY closes the week back over its 50-day with QQQ reclaiming its 10-day.

Trigger. SPY closing a day over 768.50 and Intel's report on Thursday 2026-10-22 not lowering its outlook.

Invalidated by. IWM closing under its sixty-session low of 269.88 on the same days.

Downside — 35%

Brent holds near 110, the ten-year turns up, and chip names fail to bounce on earnings. SPY loses its floor and small caps print new lows.

Trigger. Brent closing over 110 with the ten-year back over 5.15%, or any weak guidance from Lam or Intel.

Invalidated by. SPY closing a day over its 10-day near 774 on Thursday or Friday.

The week ahead

• 2026-10-21Tesla reports after the close (company 8-K; Zacks consensus 44 cents on 27.8 billion); Lam Research scheduled the same day per our cached calendar
• 2026-10-22Intel scheduled to report; the stock fell 5.8% Friday into the print
• 2026-10-28FOMC decision at the end of the October 27-28 meeting, no projections; CME priced about 52% for a hike after CPI
• 2026-10-29September personal income and outlays (PCE) per the fedratecalc release schedule
• 2026-12-09FOMC decision with the next Summary of Economic Projections

The next five sessions

This week

No scheduled macro print is large enough to reset the Fed path this week. The October jobs report and CPI both land in early November, and September PCE is dated October 29, a day after the FOMC decision, per fedratecalc. That leaves earnings as the event flow, and the order matters. Tesla and Lam Research are scheduled for Wednesday after the close and Intel for Thursday. Zacks has Tesla at 44 cents on 27.8 billion of revenue, a single-source number. Lam is the cleaner read on the build, because it sells the equipment upstream of the memory and logic names that sold off Friday. Intel lost 5.8% on the session, so its report arrives with the stock already marked down.

Chips are the hinge. The iShares Semiconductor ETF lost 4.14% Friday, per the AP-derived stockti recap, and QQQ sits 4.4% under its 10-day with a daily stochastic near the floor. Those are oversold readings. Oversold is a condition and not a signal. The thing that turns it into one is a close back over the 10-day, which is 742.88 on QQQ and about 774 on SPY, both a long way up from here.

The rate side sets the ceiling. After September CPI the October meeting moved from about 22% to about 52% on CME's tool, per Reuters on October 14. Twelve days earlier prediction markets had it at 16%, per Benzinga. A market that swings 36 points on one print is pricing the next number and not a view. The next number is oil. Brent settled at 109.86 Friday, per stockti citing AP, and gasoline supplied more than half of September's monthly CPI gain. Crude that holds here is inherited by the October report.

Options mark where the pins sit. The October 19 and 20 expiries carry their largest put open interest at 731, with 18,000 to 19,000 contracts, and calls stacked at 780 to 784 (OI as of October 15; open interest and not gamma). The 767 put strike held 10,163 contracts and now sits above spot, a former floor that is overhead.

The next four weeks

This month

Ahead this month, two scheduled policy datapoints bracket the window. The FOMC decides on October 28 without projections, and September PCE arrives the next day. Pricing is split. CME showed about 52% for an October hike after CPI, and December odds of 74.3% a month on from 50.2%, per AP-derived stockti. Goldman and Citi both put the next hike in December, and BofA is the outlier with 50 basis points more by year-end, per the Reuters factbox as carried by capital.com. A committee that raised to 3.75% to 4.00% in September and wrote in its minutes of one more increase is being priced for more than the minutes said.

PCE is the datapoint that can argue back. Cooler-than-expected PCE for July and August is part of why October odds had collapsed before CPI, per the Reuters reporting carried by Kitco. Core CPI is 2.6% on the year. A PCE print that runs under CPI again would leave the committee holding two inflation readings that disagree, which is the setting in which a hold with an open door is the easiest decision.

The earnings season is the second driver. LSEG IBES had third-quarter index growth above 30% as of last week, and the banks that reported showed the split between trading desks and everything else. Goldman's equities revenue was 6.44 billion, up 72%, and Morgan Stanley's equities line rose 41% to 5.8 billion, per Morgan Stanley's 8-K. Morgan Stanley still lost about 3% on the day, and Wells Fargo fell 1.4% on a beat, per HeyGoTrade. Good numbers that sell are the same signal as last month's: the market is paying for what is next.

What is next is 2027 capital spending. The chip break is a question about whether the build can finance itself. Hyperscaler debt issuance reached 159 billion year to date, 91% of a 175 billion forecast, per aiweekly citing Goldman, and the same Goldman data has AI-linked issuers at about 115 basis points over Treasuries against 78 for broad investment grade. The calls this month carry the guidance that answers it.

The next quarter

This quarter

Ahead through year end, the quarter turns on whether leadership repairs or the break widens. Two numbers set the stakes. BofA's October fund manager survey, released October 13, had a net 37% overweight technology, up from 30% in September and the heaviest since August 2020, with cash at 3.9%, under the 4% level the survey treats as a sell signal, per the survey coverage. The same survey named a disorderly rise in bond yields as the top tail risk and an AI bubble bursting as the next. The quarter's first real test of both arrived within days.

The positioning was one half. The financing was the other. A ten-year that eased to 5.09% did not support QQQ, and the 30-year has been above 5% for its longest stretch since 2007, per aiweekly. If rate relief does not protect the group, the group is trading on the supply of its own financing, and supply is a December and January story as new issuance clears.

The small-cap side has not recovered. IWM is under its 200-day for the first time in this run, 2.6% below, and its 50-day is 6.4% overhead. That is the opposite of a broadening setup. The December 9 meeting adds projections, and a dot set that shows the end of tightening is the one condition in which both the long end and the financing story could ease together.

The quarter's honest uncertainty is the earnings line. Strong results have not been rewarded for two weeks. If 2027 guidance holds and the stocks still do not respond, the market is telling us the multiple is the constraint. If guidance weakens, the quarter has a second problem.

The next twelve months

This year

Ahead over the year, the argument has two versions, and the ten-year's fall alongside stocks separated them. In the discount-rate version, the multiple falls when the ten-year rises. That version predicts stocks rally when yields ease, and they did not. In the financing version, the build is paid for with debt that is repriced by supply, and equity suffers when issuance is large even if the ten-year eases. That version fits last week.

The numbers on the financing side are large. Hyperscaler gross issuance of 159 billion for 2026 so far is 91% of the full-year forecast, and Goldman's forecast for next year is 420 billion, per aiweekly citing Reuters. If that is right, the bond market has to absorb nearly three times as much issuance in 2027 as it has this year, at spreads that are already wider than the broad market's.

On the street, the published year-end numbers have not moved since last month. Ed Yardeni cut his to 7,900 on September 16 by lowering his multiple, per Advisor Perspectives, and Citi sits at 8,100. Neither figure has been changed since the break, and the cash index now trades under both by a wider margin than it did a week ago. A strategist who cut because yields rose is now facing yields that fell with stocks falling.

The tests are unchanged. A ten-year under 4.80% with 2027 estimates intact moves the year to constructive. Two months of downward 2027 revisions moves it to defensive. Until one of them prints, the year is a bet on which version of the argument is right.

The levels

Into the week SPY sits at 763.25, under its 10, 20 and 50-day averages, with the 200-day at 725.01. QQQ sits at 709.80 near its 20-day low of 707.23. IWM, at 270.44, printed a sixty-session low of 269.88 and trades 2.6% under its 200-day.

• SPYThe 50-day is the first reclaim and the line that separates a shakeout from a range change. The 731 put strike carries the largest open interest for the October 19 and 20 expiries, and 725.01 is the 200-day and the quarter's graded line.
• QQQQQQ closed on a 20-day low with a daily stochastic of 4.6. The 50-day is the first line a bounce has to clear. The 10-day and 20-day are 4.4% overhead, which is the distance an oversold reading has to cover to matter.
• IWMIWM sits under its 200-day and under its 10-day, with the 50-day 6.4% overhead and a daily stochastic of 3.8. A close under the sixty-session low is the level that says the credit channel is widening.

What would change it

Watch three things. Whether SPY closes a day back over its 50-day. Whether Brent holds over 105 as the ten-year turns back up. And whether IWM closes under 269.88 while the leaders are still oversold.

Leadership

• Leading: Energy producers, helped by Brent's 3.7% Friday rise, Long Treasuries, with TLT up 2.65% on the week, Banks with large trading desks, with Goldman and Morgan Stanley equities revenue up 72% and 41%
• Lagging: Semiconductors, with the iShares Semiconductor ETF down 4.14% Friday, The Nasdaq-100 proxy, down 5.52% on the week, Small caps, six weeks down and under their 200-day, Fuel-heavy operators, as crude is up 9% on the week by the USO proxy

The line has moved. Last week it ran between suppliers to the AI build and borrowers. This week the suppliers broke too, and the winners are those who sell or own the commodity and the desks that trade the volatility.

Cross-asset

• oil (oil; USO is a WTI futures proxy and not WTI itself). USO gained 9.15% on the week and 11.05% on the month, now 13.4% over its 50-day. Brent settled at 109.86 Friday, per stockti. Gasoline was more than half of September's CPI gain.
• long\_treasuries (20y+ Treasuries, TLT as the vehicle). TLT gained 2.65% on the week and sits 0.8% under its 50-day. Duration rallied while QQQ lost 5.5%, which is the week's cross-asset oddity.
• dollar (US dollar; UUP is a DXY proxy and not DXY itself). UUP fell 0.34% on the week but is up 0.52% on the month and 3.9% over its 200-day. A firm dollar keeps financial conditions tight for the build.
• gold (gold; GLD is the ETF proxy). GLD fell 0.67% on the week and sits 7.9% under its 200-day. Gold did not take the haven bid that the equity break might suggest.
• volatility (VIX futures ETF, not spot VIX). VIXY gained 8.77% on the week yet sits 26.5% under its 200-day. The selling has been orderly so far; a break above its 50-day would say that has changed.

The record

Last Sunday's note set one test: SPY's Friday close inside 767.57 to 790.00. It closed at 763.25. Grade: wrong. A hot CPI and a chip-led break took the index under its 50-day. Week horizon: 4 right, 1 wrong, 2 mixed over 7.

• This week — wrong. We said: Neutral at 0.50 on one test: SPY's Friday 2026-10-16 close inside 767.57 to 790.00, with September CPI and the first six banks as the swing variables and a 3.7% headline print as the downside trigger. What happened: SPY closed Friday at 763.25, 4.32 points under the 767.57 floor of our band, after losing 1.97% on the week. Headline CPI printed 3.7% against the 3.6% poll, the downside trigger we named, and a chip-led break took the index through its 50-day.
• This month — too early to call. We said: Constructive at 0.55, wrong if any Friday close through 2026-10-30 printed SPY under 749.60, right on a Friday close above 790.00. What happened: The first two Fridays closed at 778.57 and 763.25. The second sits 13.65 points above the 749.60 floor and 26.75 under the 790.00 line, so neither trigger fired.
• This quarter — too early to call. We said: Neutral at 0.55, wrong to the upside on two Friday IWM closes above its 50-day with SPY over its high, wrong to the downside on a Friday SPY close under its 200-day or two months of falling 2027 estimates. Graded 2026-12-31. What happened: IWM closed at 270.44 for a sixth straight weekly decline, 6.4% under its 50-day. SPY stayed 5.3% over its 200-day. Neither trigger fired.
• This year — too early to call. We said: Neutral at 0.50, back to constructive if the ten-year returned under 4.80% with 2027 estimates intact. What happened: The ten-year eased to 5.09% from near 5.25%, per AP, still 29 basis points above the 4.80% line. LSEG IBES growth for the third quarter stayed above 30% as of the last read.

Week horizon: 4 right, 1 wrong, 2 mixed over 7 graded notes. Month, quarter and year: 0 graded, 21 too\_early across those same 7 notes. The archive holds 7 prior notes, dated 2026-08-30, 2026-09-06, 2026-09-13, 2026-09-20, 2026-09-27, 2026-10-04 and 2026-10-11; the first had nothing to grade.

Sources

• stockti, 2026-10-16, semiconductor selloff sinks Nasdaq 2.1% as weekly tech slide deepens (AP-derived)
• AJC, 2026-10-16, US markets weighed down by extended sell-off in the tech sector (AP)
• Jamaica Gleaner, 2026-10-18, US markets weighed down by an extended sell-off in tech sector
• Reuters via Kitco, 2026-10-14, US inflation heats up in September; Fed rate hike this month back on table
• Reuters via Kitco, 2026-10-14, Fed rate hike call seen tight as inflation heats up
• XTB, 2026-10-14, US inflation concerns rise
• Reuters factbox via capital.com, mid-October 2026, major brokerages expect next Fed rate hike in December
• investinglive, 2026-10, Goldman Sachs moves its call for the next Fed rate hike to December from October
• Benzinga, 2026-10-02, via search: "Fed Hike Odds Collapse to 16% on Prediction Markets After Soft Jobs Report"
• BofA Global Fund Manager Survey, October 2026, via Crypto Briefing
• XTB, 2026-10-13, US500 near record highs; weaker BofA survey and NFIB data fail to dent sentiment
• aiweekly, 2026-10, Alphabet, SpaceX equity surge renews AI debt-binge fears
• Morgan Stanley, 8-K, third quarter 2026 earnings release
• HeyGoTrade, 2026-10, bank stocks news: Q3 2026 earnings and the Fed rate hike
• Tesla, 8-K, third quarter 2026 earnings date
• straits.live, 2026-10-08 daily brief on Brent and Hormuz transits
• fedratecalc, October 2026 monthly release schedule
• Advisor Perspectives, 2026-09-16, Yardeni cuts S&P 500 to 7,900 (cited in prior issues)

Prices, ranges, moving averages and open interest in this note come from our own market-data bars through 2026-10-16. Every external claim names its source above. This note is forward-looking market research: it describes scenarios and the conditions that would confirm or invalidate them. It is not investment advice, not a recommendation, and states no forecast, price target or trade.