Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Run the checklist on the third quarter of 2026 and you would expect a bloodbath. Brent crude pushed back above $100 a barrel. The benchmark US government bond yield at levels not seen since the dot com hangover. A Federal Reserve that has stopped cutting and started hiking again. A war in the Middle East with no settlement in sight. And a growing chorus asking whether the AI spending spree can keep running at this pace.
Instead the S&P 500 finished the quarter in the green, and the Nasdaq has printed fresh records since. For the casual observer, that looks like strength. For a contrarian, it looks like a question that deserves a proper answer: what exactly is holding this market up, and how much weight can that support carry?
The short version is that earnings are doing nearly all the work, a small group of companies is doing most of the earnings, and the bond market is quietly raising the bar every stock has to clear. None of that is a crash signal on its own. All of it is a reason to stop treating index highs as proof that the coast is clear.
The Headwinds Are Not Imaginary
Start with the bond market, because it is the story most equity investors are underweighting. On 1 October the US 10-year Treasury yield touched 5.342%, the highest reading since early 2002 and above its 2007 peak. The move into that level was violent. Over the three months to September the 10-year added roughly 87 basis points, the sharpest quarterly climb since early 1994, and the 30-year yield pushed through 5.65% alongside it.
The drivers are not mysterious. Bloomberg’s account of the selloff points to persistent inflation, heavy government borrowing and resilient growth, with an extra twist that matters for this article: the scale of AI infrastructure investment is itself competing for capital and nudging borrowing costs higher worldwide. The AI boom that props up equity earnings is also one of the forces lifting the discount rate applied to those earnings.
[Also Read: How To Profit From the Next Market Crash]
Then there is energy. With the Middle East conflict still unresolved, Brent has been trading back above $100 a barrel, while US crude sits below that mark, a spread that points to stress in seaborne supply rather than anything happening in American oilfields. The US Energy Information Administration estimates global inventories have shrunk by around 400 million barrels so far this year and expects the drawdown to continue through year end. Oil is not just a pump price. It runs through freight, airlines, plastics, fertiliser and food, which means it eventually runs through inflation.
We have argued before that the oil market is telling two different stories, and that one of them is wrong. The gap between paper and physical crude has not closed, and the equity market is still pricing the friendlier of the two.
And inflation is where the Federal Reserve comes in. The Fed raised rates by a quarter point in September, its first hike in roughly three years. The August personal consumption expenditures report offered some relief: headline PCE came in at 3.4% year on year and core at 3.0%, both under forecasts. Futures traders responded by pushing their expected next hike from October towards December. But 3% core is still a full percentage point above target, and as one economist noted, the August numbers predate September’s spike in diesel costs. Relief is not resolution.
Profits Are Doing the Heavy Lifting
If the macro picture is so hostile, why has the index held up? The honest answer is that corporate profits have been growing fast enough to outrun it. FactSet’s latest preview has analysts pencilling in year-on-year earnings growth of 29.5% for the third quarter and 32.4% for calendar 2026. If the quarterly figure lands, it will be the third quarter in a row above 25%. Those are numbers you normally only see coming out of a recession, not in the middle of an expansion with tight monetary policy.
The second quarter was even stronger. FactSet’s July update had the blended growth rate near 38%, the best since 2021, but it also carried a detail that every bull should read twice: strip out Alphabet alone and that figure drops to roughly 26%. One company was responsible for about a third of the headline growth rate. That is not a broad based profit boom. That is a profit boom with a very heavy keel.
The upside of fast earnings growth is that it has quietly deflated valuations. The forward price to earnings multiple on the S&P 500 started the year at about 22 and now sits around 19, close to its ten year average, even though the index itself is up double digits. Price went up. Earnings went up faster. On the surface, that is the healthiest possible way for a market to climb.
But a contrarian reads that same compression differently. A multiple can fall for two reasons: because profits are surging, or because investors are getting nervous about paying for them. Goldman Sachs suggests both are at work, citing higher rates and growing doubt over whether AI driven profits are sustainable as the reasons the multiple has come down. In other words, the market is still buying the earnings, but it is already applying a discount for the possibility that they do not last.
An Index Standing on Fewer Legs
Here is the part that index level headlines hide. Goldman Sachs strategist Ben Snider told clients in late September that the S&P 500 sits within about 1% of its record while the median stock in the index is roughly 16% below its own 52-week high. By the bank’s preferred gauge, that gap is the widest since the dot com bubble. The index looks like a bull market. The typical constituent looks like a correction.
The engine behind that gap is the AI build out. The largest US hyperscalers are on course to spend around $800 billion on capital projects this year, roughly double last year’s outlay. Wall Street consensus has 2027 at about $1.1 trillion, and Goldman now expects closer to $1.2 trillion, with the spenders needing around $300 billion a year in AI revenue just to break even. That money flows to chipmakers, memory suppliers, server builders, power equipment firms, cooling specialists and network vendors, which is exactly why so much of the index’s profit growth is tied to one theme.
When breadth gets this thin, the index stops being a diversified bet on the American economy and becomes a leveraged bet on a narrow set of companies all exposed to the same capital cycle. That cycle has a built in vulnerability: one company’s capex is another company’s revenue. If the spenders ever slow, the revenue disappears from the suppliers at the same moment the spenders’ own depreciation bills arrive. Snider himself frames the risk with care. Narrow breadth has historically flagged drawdown risk, but he also sees room for laggards to catch up if the macro fog lifts.
That is the right way to hold the idea. Narrow breadth is not a sell signal. It is a fragility signal. It tells you the market is more sensitive than usual to one specific disappointment.
The Bond Market Has Become a Real Competitor
For most of the last fifteen years, owning equities was close to a default choice because the alternative paid almost nothing. That excuse is gone. A forward multiple of 19 implies an earnings yield of roughly 5.3%. A 10-year Treasury now pays about the same, with the full faith and credit of the US government behind it and none of the drawdown risk. On the crudest measure, investors are being paid almost nothing extra to hold the riskier asset.
The picture improves when you measure it properly, and that is worth being fair about. Goldman’s own work compares the S&P 500 earnings yield with the inflation adjusted 10-year yield and finds a spread of around 270 basis points that has been fairly stable for two years. Equities still carry a premium. It is just thinner than the long run norm.
Fortune ran the more conservative version of the same arithmetic using trailing GAAP profits and TIPS yields, and arrived at a cushion of only about one percentage point. Whichever method you prefer, the direction is the same. The margin of safety that used to justify owning stocks through rough patches has narrowed, and every further rise in yields narrows it more.
The Goldman research adds one more detail that matters here. It estimates that around three quarters of the S&P 500’s present value comes from cash flows ten or more years out. That is why long dated yields bite equities hardest, and why the most richly valued growth names, whose profits sit furthest in the future, are the most exposed if the bond selloff resumes. For a contrarian, this is the variable worth watching more closely than any single earnings report.
Four Roads From Here
Nobody knows which way this resolves, including the strategists quoted above. The useful exercise is not to pick a winner but to sketch the plausible paths and ask what would hold up on each one.
| Scenario | What happens | What tends to hold up |
|---|---|---|
| AI keeps compounding | Capex keeps rising, earnings beat, inflation drifts lower, yields ease | Broad equity exposure and growth names |
| Sticky inflation | Oil stays high, the Fed hikes again, yields stay above 5% | Value, dividend payers, short duration bonds |
| Recession | Consumers retrench, unemployment rises, profit estimates are cut | High grade bonds and cash |
| Sharp equity sell off | The AI trade unwinds and narrow leadership drags the index lower | Dry powder, and the patience to use it |
The last row deserves a calmer reading than it usually gets. Sell offs are not rare events. J.P. Morgan Asset Management’s long running data shows the S&P 500 has suffered an average peak to trough drop of about 14% within a calendar year going back to 1980, yet most of those years still finished higher. A double digit pullback is the normal cost of admission, not evidence that something has broken.
For an investor who does not need the money for years, a sharp fall is mainly a pricing event. For an investor who does need it soon, the same fall can force a sale at the worst moment. The scenario itself matters less than which of those two investors you are.
A Portfolio That Does Not Need a Forecast
If you cannot know which road the market takes, the sensible response is a portfolio that does not depend on guessing right. That sounds obvious, yet most of the damage investors do to themselves comes from quietly building around a single prediction: that rates will fall, that AI will keep winning, that the crash is coming next month.
A forecast free structure for someone still building wealth tends to share a recognisable shape. The core, often somewhere between three fifths and four fifths of the total, sits in broad, low cost index exposure that owns the whole market rather than a theme. Around that core sits a layer of deliberate diversifiers, perhaps a tenth to a quarter, chosen precisely because they do not move in lockstep with mega cap technology: international developed and emerging markets, smaller companies, value and dividend focused holdings. The remainder, anywhere from a tenth to nearly a third depending on age and nerve, goes into high quality bonds and cash.
The ranges are wide on purpose. A thirty year old with stable income and decades of contributions ahead can sit at the aggressive end. Someone five years from retirement, or already drawing an income, belongs at the cautious end, with a meaningful cash buffer, short and intermediate bonds, and an equity sleeve tilted towards established, cash generative businesses.
Notice what this structure does in the current environment. It still participates if the AI cycle keeps running, because the core index is dominated by those very companies. But the diversifier layer is an explicit hedge against breadth staying narrow and then snapping, and with Treasuries paying above 5%, the bond sleeve is no longer dead weight. For the first time in a long while, the defensive part of a portfolio actually pays you to hold it.
Individual stock picks, including AI leaders, can sit on top of this framework for those who enjoy the research. The point is that they sit on top. They should never be the foundation, because a concentrated position demands answers to questions most investors never settle in advance: when to add, when to trim, and what would make you sell.
Three Stress Tests Worth Running Tonight
Before deciding whether your portfolio is too brave or too timid for this market, put it through three blunt questions.
- If equities dropped 30% next week, would you be forced to sell any of them to cover living costs? If the honest answer is yes, the problem is not your stock picks. It is the size of your cash and bond buffer.
- If equities rose 30% instead, would you be pleased or furious? Furious means you have drifted too defensive out of fear, and that fear has a measurable cost when markets keep compounding.
- If your strongest conviction about the economy turns out to be wrong, does the plan still work? This is the one that matters most, and the one most investors skip. A plan that only succeeds if rates fall, or only if AI keeps booming, is a bet wearing a plan’s clothes.
The market’s resilience this year is real, and it rests on something real: corporate profits growing at a pace rarely seen outside a recovery. But resilience built on a narrowing base, priced against the highest risk free yields in a generation, is not the same thing as safety. The contrarian position here is not to call a crash. It is to refuse to let a record high on the index talk you out of preparing for the scenarios it is currently ignoring.
Own productive assets. Spread them wider than the index does on its own. Keep enough dry powder that a bad quarter is an opportunity rather than an emergency. Then let the forecasters argue about the Fed.
