Ninety Days From Today

This is the fourth installment of a series I started on September 3. Same calendar discipline, same premise, new window. Ninety days from this morning is January 30, a Saturday.

The furniture in the room first. The Federal Reserve meets twice between here and there, on December 8 and 9, which comes with a fresh Summary of Economic Projections, and on January 26 and 27, which does not. In between sit a midterm election on November 3, two days from now; the Treasury’s quarterly refunding on November 4; a White House hearing on the removal of a Fed governor on November 5; October employment on November 6; October inflation on November 10; Nvidia’s fiscal third quarter on November 17; a Thanksgiving on November 26; November employment on December 4; November inflation on December 10; a government funding deadline on December 11; a Christmas on December 25; a new Congress on January 3; and the December employment and inflation reports in the second week of January before the Fed meets on the twenty-sixth and twenty-seventh.

Three things changed in October, and the first one changes the whole argument.

The index made a new high and then broke. The S&P 500 closed at 7,818.93 on October 6, its twenty-eighth record close of the year and its first since August 13, with an intraday peak of 7,844.52 (Morningstar). By October 30 it had given back all of that and more, closing at 7,580.07, down 3.05% from the record (Morningstar).

The crack did not come from the war. It came from the sector that had been carrying the index. In three weeks the market repriced the AI trade twice: once on October 22, when Meta fell 7% after raising capital spending and posting its first cash outflow in a decade, and again on October 29 and 30, when Alphabet lost about 12% on AI-damaged search revenue and Intel fell 28.6% on guidance that missed badly (AP; CNBC).

And the discount rate finally moved in the index’s favor. The ten-year Treasury fell from 5.29% at the end of September to 4.94% at the end of October, thirty-five basis points in a month, which changes the arithmetic on every long-duration asset in the index (U.S. Treasury).

Read those three together and you have the question of the next ninety days. The market lost confidence in its load-bearing sector at the same moment the cost of capital fell. One of those should have been enough to hold the index up, and it was not.

Where the Market Stands Today

The numbers as of the October 30 close.

  • The S&P 500 closed at 7,580.07, down 1.82% on the day, down 0.93% for October, up 10.73% for the year, and 3.05% below its record close of 7,818.93 set October 6. It remains 19.49% above its 2026 low of 6,343.72, set March 30 (Morningstar).
  • The Nasdaq Composite closed at 25,437, down 2.86% on the day, its worst session since June, and the Dow closed at 51,770.31, down 1.09% but still up 1.5% for the month (AP).
  • The Nasdaq 100 fell 3.4% on the final session and sits close to correction territory, ten percent below its last record. The Nasdaq Composite ended October about 7.8% below its October 6 record (NBC News).
  • The VIX closed at 19.42, up from 17.20 the day before, after trading as high as 23.93 intraday (CBOE).
  • The forward 12-month P/E is 18.7, below the five-year average of 19.8 and below the ten-year average of 19.1. The trailing multiple is 25.9, above both (FactSet).
  • Earnings are still doing the work. The forward multiple has fallen from 20.4 at the end of June to 18.7 today while the index is up 10.73% for the year, because estimates grew faster than prices. Third-quarter blended earnings growth is 30.2% with revenue growth of 13.4%, and 82% of companies have beaten on earnings (FactSet).
  • But the size of the beats is shrinking. The aggregate earnings surprise is 1.9%, the lowest since the fourth quarter of 2022, against a five-year average near 7.5% (FactSet; Bloomberg via Investing.com).
  • The fourth quarter is expected to grow 28.4%, and for the first time since 2022 the Magnificent 7 are expected to trail the other 493 companies, at 22.6% against 30.8% (FactSet).
  • The bottom-up target price is not the same as the strategist consensus. The Bloomberg survey of nineteen strategists, published October 26, put the average year-end 2026 target at 7,967, down just nine points from September, which implies about 4.5% upside from the October 23 close (Bloomberg via Investing.com).
  • Goldman Sachs keeps a year-end 2026 target of 8,000 and a 2027 target of 8,700, and UBS has a mid-2027 target of 8,400 (Yahoo Finance; InvestingLive).
  • The 10-year Treasury is 4.94% and the 30-year is 5.31%. The rolling correlation between oil and the ten-year, which ran near 0.96 for most of this year, has broken down in the last two weeks of October (U.S. Treasury; CNBC).
  • The consumer keeps deteriorating. The University of Michigan’s final October sentiment index is 48.0, the second-lowest on record, with year-ahead inflation expectations at 4.7% (University of Michigan).
  • Gasoline is $4.20 a gallon and diesel is $6.32, with the mortgage rate at 7.24% (AAA; Freddie Mac).

Put the fifth bullet and the twelfth side by side. The index trades at 18.7 times forward earnings, a discount to its own five-year average, and the consumer who buys the earnings is at the second-lowest sentiment reading ever recorded. Both of those are true, and the distance between them is what the next ninety days will resolve.

The Wall Cracked

Last month this section quoted Invesco’s Brian Levitt saying the rally “is going to end when something breaks in the AI trade,” and I called it a statement about a building with one load-bearing wall (CNBC).

In October the wall cracked, and the interesting part is how little it took.

The sequence began on October 22 with Meta. The company beat on revenue. It raised its capital spending forecast and reported its first cash outflow in more than a decade, and the stock fell 7% on a day when the Dow finished fractionally higher (AP). That is not a company in trouble. That is a company whose shareholders did the arithmetic on what the spending is for and did not like the answer.

A week later the arithmetic got louder. Alphabet fell about 12% on October 29, losing roughly $500 billion of market value, its worst session on record, after disclosing that AI-generated search summaries were hurting its core advertising business while guiding to as much as $205 billion of capital spending this year (CNBC). The next morning Intel fell 28.6%, its worst day since the 1970s, on guidance of 25 cents a share against an expected 43, and Amazon held a $220 billion capital budget while running negative free cash flow for a second consecutive quarter (BBC; CNBC).

Add those up and you get a number that should be printed on every prospectus in the sector. Meta, Alphabet, Amazon, Microsoft, and Oracle have committed roughly $800 billion to capital spending in 2026, and big technology’s total AI outlay is approaching $1 trillion (CNBC). That money has to come from somewhere, and the somewhere is a combination of operating cash flow that is now going negative at two of them, debt issued into a ten-year that spent the summer climbing, and equity sold to investors who have just discovered they can reprice the story in an afternoon.

This is the mechanism the index has been hiding behind all year. It is not that AI earnings are fake. It is that AI earnings and AI spending are the same number viewed from two ends, and the market had been valuing the first while ignoring the financing of the second. The financing cost rose all summer, from 4.44% on the ten-year at the end of June to 5.29% at the end of September. October is when the market finally repriced the other end.

The good news is that the index absorbed it. A 3.05% drawdown from a record, a VIX that closed at 19.42, and an equal-weight index that is still 13.89% ahead for the year is not a market in panic (Advisor Perspectives). Breadth held. Money rotated into energy, staples, and financials rather than leaving.

The bad news is what the crack revealed about the keystone. The S&P 500 gained 2.03% in the third quarter while the Dow lost 2.7% and the gains were concentrated in a handful of names. That means the index’s return was never the economy’s return. It was the return of a handful of companies financing a buildout, and in October those companies were asked to show the invoice.

The Math That Changed Sign

Here is the part that ought to make a careful investor uncomfortable, because it cuts against both the bulls and the bears.

Start with the earnings yield. An index at 18.7 times forward earnings yields 5.35% in earnings. The ten-year Treasury pays 4.94%. The difference is about forty-one basis points.

That is the price of owning the entire equity market in place of a government bond. Forty-one basis points, for the risk that earnings disappoint, that multiples compress, that a war closes a strait, and that a Congress refuses to fund a government. A gap that thin is not a valuation call. It is a statement of confidence, and confidence is the one input this article cannot source.

You can argue the earnings are understated, and FactSet’s numbers give you ammunition: 32.9% earnings growth for calendar 2026 and 16.3% for 2027, with a fourth quarter expected to grow 28.4% (FactSet). A 16.3% growth rate on a 5.35% earnings yield is a very different proposition from a static one.

But two things have to go right for that arithmetic to pay. The earnings have to arrive, and the discount rate has to stay where it is or fall. In October the second condition improved by thirty-five basis points. The first condition is now the entire question, and it is concentrated in a company that reports on November 17.

Nvidia’s guidance for its fiscal third quarter is $108 billion in revenue at a gross margin near 74% (Nvidia investor relations). Its prior quarter grew revenue 106% year over year. If that continues, the earnings side of the equation holds and the strategist consensus of 7,967 proves too low. If it does not, the market has to reprice every company that was valued on the assumption the buildout continues, and it will do it in one session.

The Fed’s role in this is smaller than it looks. A December hike is priced at roughly 85% and would take the funds rate to 4% to 4.25% (Reuters via Investing.com). Against a 4.94% ten-year that is already in the price, a quarter-point move in the policy rate is not the event. The event is what the hike says about 2027, and the Fed’s own projections have the median participant at 4.1% at the end of 2026, which implies one move and a long pause (Federal Reserve SEP).

What changed in October is more subtle and more important than the hike. For nine months this index was priced against a rising long bond, and every escalation in the Gulf reached the valuation model twice: once through the energy earnings inflating the growth rate and once through the discount rate deflating everything else. That double exposure is gone. Oil rose 13.6% in October and the ten-year fell. The index now faces a single, cleaner question, which is whether the AI earnings are as durable as the multiple assumes. It is a harder question than the war, because the war had an obvious answer and this one does not.

The Corruption Premium

The standing question in this series is whether the quality of the government writing the rules has started showing up in the cost of capital. October supplied the largest test case yet, and it is not a scandal. It is an audit request.

The tariff refund is now a distribution of roughly $166 billion made by filing date. The Supreme Court held in February that the emergency statute does not authorize the tariffs. Customs collected the money on about 53 million entry summaries from roughly 330,000 importers, and about $100 billion had gone back by August (Representative Jill Tokuda). On October 21, Senator Raphael Warnock, Representative Tokuda, and Representative Brad Sherman asked the Government Accountability Office to audit whether the refunds are “complete, accurate, properly authorized, and protected against improper or duplicate payments,” which importers received them, and whether any of the benefit reached the consumers who paid the surcharge (Senator Raphael Warnock). On October 8, Judge Richard Eaton certified an injunctive class covering every importer of record who paid the duties, opening a path for companies that never sued, while the Department of Justice continues to argue that relief belongs only to the named plaintiffs (ArentFox Schiff).

For an equity holder the relevant sentence is not about fairness. It is about predictability. Two importers with identical claims, identical harm, and identical paperwork will receive different outcomes because one filed and one did not, and nobody can tell you today which of the hundred and sixty-six billion dollars in contested refunds will be resolved by a rule and which by a lawyer. A market prices that uncertainty in the only place it can, which is the discount rate on the companies exposed to it.

The exemption channel has moved to pharmaceuticals, and it is larger. Under Proclamation 11020, patented drugs from most countries face duties of up to 100%, while nine specialty categories, including orphan drugs, gene therapies, antibody-drug conjugates, and plasma-derived therapies, pay 0% if the product originates in one of nineteen named jurisdictions or if Commerce grants an “urgent health need” determination case by case (Federal Register; WilmerHale). The categories make clinical sense. The discretion is the part that has a price, and the Journal of Financial and Quantitative Analysis has already measured what that price looks like in the older tariff exemption program: contributions to the party in power raised the odds of approval, and each grant was worth a median of about $51 million in firm value (JFQA).

And the pressure on the Fed has a courtroom and a date. The White House has convened a committee of its own appointees to examine Governor Lisa Cook’s mortgage statements, with a hearing set for November 5, two days after the midterms (CNN). The Fed held rates on October 28 by 10 to 2, and the two dissents were for a hike, one of them from Cook herself (Federal Reserve statement). A removal proceeding aimed at the board’s least dovish member is not a case about a mortgage. It is a case about who sets the price of money, and the market has to price the possibility that the answer changes.

The index at 18.7 times forward earnings has not decided what to charge for any of this. It is charging forty-one basis points over the risk-free rate for the whole market, and that number embeds an assumption that the rules stay stable enough that nobody has to think about them. Every item in this section is a reason that assumption now carries more weight than it once did.

The Ninety-Day Map

Take the premise seriously: no improvement in the war, no breakthrough, no formal reopening of the strait. Here is the path from November 1 to January 30.

November 3 through November 10. The midterms, the refunding, the Cook hearing, October employment, and October inflation, all inside eight days. Forecasters have Democrats heavily favored in the House and favored in the Senate, and the historical pattern after a midterm is one of the strongest twelve-month stretches in the cycle. That pattern assumes the uncertainty resolves. A contested count, a shutdown threat in a lame-duck session, or a payroll report of another 29,000 jobs would each break the assumption.

November 17. Nvidia. This is the single dated event that decides the year, and it lands four weeks before the Fed meets. Guidance is $108 billion in revenue at a 74% gross margin. The prior quarter grew 106%. Anything close to that and the melt-up case below stops being a branch.

November 26 through December 3. Thanksgiving, then the lull. The household bill arrives with diesel at $6.32 and gasoline at $4.20, and the retail guidance issued in this stretch will tell you what the companies think the consumer can still absorb.

December 4 through December 11. The cluster. November employment on the fourth, the Fed on the eighth and ninth with a fresh dot plot, November inflation on the tenth, and the continuing resolution expiring on the eleventh (Congress.gov). A rate hike, an inflation print, and a funding cliff in seventy-two hours.

January 3 through January 30. A new Congress, the December employment and inflation reports, and a Fed meeting with no projections. This is the stretch that decides whether October’s de-rating was a repricing or the beginning of a bear market, and until the Fed meets it has no scheduled event in it. The absence of a catalyst is itself the catalyst, because a market with nothing ahead looks backward.

That is the base case, and it lands the index between 7,200 and 7,850: a market that has given back the October record, kept most of the year’s gain, and spent the quarter discovering that its largest sector has a price.

The escalation branch. A full closure of the strait, or strikes that draw Iran’s neighbors in directly, is the 1973 scenario. From its January 11, 1973 peak to its October 3, 1974 trough, the S&P 500 fell 48.2% over 630 days, and the round trip to the old high took more than nine years (History of Market). Measured from the October 6 record, that is roughly 4,050. The Volcker precedent is the more instructive one, because it is about what a central bank does when a supply shock refuses to leave: from November 28, 1980 to August 12, 1982, with the funds rate pushed toward 20%, the index fell 27.1%, which from here is roughly 5,700 (Forbes). And the milder analog, if the shock stays contained: in 1990 the index fell 19.9% from its July peak to October 11, which from here is roughly 6,260 (A Wealth of Common Sense).

The paradox branch. The earnings are real, the multiple is below its five-year average, and the discount rate just fell thirty-five basis points. If Nvidia confirms guidance, the fourth quarter delivers something close to the 28.4% the analysts expect, and the Fed holds in December on a soft payroll print, then the market has a reason to buy that has nothing to do with the war and everything to do with arithmetic. A close above 8,300 before January 30 is not a fantasy in that world. It is what happens when a de-rating reverses and the earnings never showed up in the denominator to justify it.

The point of the map is not that the bear case is certain. It is that the base case now depends on a company reporting on a single day in November, while the melt-up depends on that company and the Fed and the consumer all behaving at once.

What a Careful Investor Does

The playbook rhymes with the bond article’s, because it is the same storm seen through a different window.

  • Do not sell the core; rebalance the concentration. The index sits 3.05% below a record set four weeks ago, and the sector that carried it has already told you it can lose a trillion dollars of value in three weeks. Rebalancing into strength is not timing. It is the only free lunch in a market where the largest seven companies are more than a third of the index.
  • Know which side of the earnings split you own. For the first time since 2022, the Magnificent 7 are expected to grow earnings more slowly than the other 493, at 22.6% against 30.8%. If that forecast is right, the index’s leadership has to broaden, and the money that was in the crowded trade has to go somewhere. That is where the opportunity in this window lives.
  • Own the businesses that get paid now. A 41 basis point earnings-yield premium over the ten-year is not compensation for duration risk in a company whose cash flows arrive in 2031. It is compensation for owning a company with current cash flow, pricing power, and modest debt. Energy, at 13.2 times forward earnings, is the cheapest sector in the index and the one that pays you if this article’s premise is right.
  • Treat Nvidia as a position, not a religion. Whatever you own, size it so that a 20% gap on November 17 does not change your plan. That is the only way to hold a concentrated market without being held by it.
  • Automate the buying and keep the cash earning. Dollar-cost averaging is the house recommendation for the same reason it was last month: you cannot time the bottom, but you can schedule it out of existence. And cash held for the paradox branch should be earning 4% while it waits, because a market that just showed you its seam is a market that may offer a better entry.

The Long Way of Saying It

The S&P 500 is not priced stupidly, and that is what makes this quarter dangerous. At 18.7 times forward earnings it trades below its own five-year average, because earnings grew faster than prices for four straight quarters. The bears have to argue against that number, and the number is stubborn.

What the index is priced for is narrower than the multiple suggests. It is priced for one company’s guidance on November 17, one discount rate that just fell and may fall further, one consumer at the second-lowest sentiment reading on record who keeps spending anyway, and one set of rules stable enough that nobody has to think about them.

October took a crack at the keystone and the arch held. That is genuinely reassuring, and it is also the reason to be careful. An arch that has been tested once is not stronger for it. It is simply an arch whose seam you have now seen, and the next load will find the same stone.

Storms do not send invitations, but they do keep appointments. This window has one on November 17, one in the second week of December, and a long quiet January in which the calendar offers nothing at all. Build the defenses before the lightning strikes, and remember that the stone that carries the roof is the one worth watching.

Sources

PRH | huffmanwrites.org | © Philip Huffman