Follow the leader

S&P 500 and Nasdaq 100 futures are modestly lower ahead of the open to start the week.

On Friday, a weaker than anticipated non-farm payrolls report showing US job growth of 29,000 with the unemployment rate edging up to 4.2% couldn’t reverse any recent market trends — even the bludgeoning of the bond market.

While odds of a US interest rate increase winnowed further on the data, yields across the US Treasury curve ended the day higher! The Nasdaq 100 was the only major index to close at a record.

The first day of the year was the biggest gain for semis over software ever (at the time, at least: VanEck Semiconductor ETF has had four even better days versus iShares Expanded Tech Software ETF since then!). The start of Q3 brought back-to-back sessions of iShares MSCI USA Value Factor ETF outperforming iShares MSCI USA Momentum Factor ETF by two percentage points. But we haven’t yet seen much in the way of a change of market character in the early days of Q4.

Robinhood traders modestly divested single stocks on Friday, with many popular tech names seeing the most selling pressure. And although retail traders have a penchant for buying the dip, that did not extend to Nike after the stock slid to its lowest level since 2013 in the wake of its earnings report.

As a subsidiary of Robinhood, Sherwood Media is restricted from writing about any company in which Robinhood is or was a selling group member of the IPO during the regulatory "quiet period" for that company.

A low-energy record

On Friday, Nvidia set a fresh intraday high for the first time since May — snapping one of its longest streaks of the AI boom without reaching such a milestone — but didn’t manage to post a new closing peak.

The stock has enjoyed accelerating annual earnings and revenue revisions through most of 2026, but only very sporadic periods of strong performance.

That’s in keeping with its history during the AI boom:

Shares skyrocketed into a new range after its May 2023 earnings report that unofficially kicked off the AI boom, but traded in a (split-adjusted) ~$40 to $50 for the next six months.

The stock then nearly doubled in two months before trading sideways for a quarter, before gaining about 50%. Nvidia settled into a (wide) range of $100 to $150 for a full year before it ramped higher at the end of Q2 2025 as part of the pan-market rally after US President Donald Trump watered down reciprocal tariffs.

This new range roughly held until the S&P 500’s 2026 low in March, after which shares once again exploded higher in short order.

At the risk of oversimplifying, you could say that over the past two years, once the market passes a big test, Nvidia finds a new, higher range.

“If you take a step back, though, a chart of the two so far this year shows that while they haven’t moved in unison, there is some evidence of NVDA leading and the SOX following,” wrote analysts at Bespoke Investment Group. “As semis ran in the spring and sold off in the summer, NVDA led the charge on both the way up and the way down. The stock traded sideways after that sell-off for a little over a month before regaining momentum in late July/early August. Semis, meanwhile, didn’t bottom out until early August and traded sideways through the middle of September before starting to pick up steam again in the back half of the month.”

“NVDA is interesting because sometimes it shows us a bearish divergence at important price tops. That includes the one at the prior all time high for the NDX made back in June,” writes Tom McClellan of the McClellan Market Report. “Divergences like those can be important, but sometimes frustrating. A divergence is a condition, not a signal, and you don’t know when the condition is going to matter, or whether it might get rehabilitated. But a non-divergence like we see right now is much more important, and more reliable. It says there is no problem.”

As a side note, however, Nvidia’s fresh high comes without much enthusiasm for significant upside to come this time around.

The implied volatility of 10-delta call options that expire in six months is at its lowest ever point during the AI boom.

The same trend, though not as extreme, can be seen across a host of other AI bottleneck stocks, like Micron, Intel, Marvell Technology, Bloom Energy, and Lumentum. The decrease in costliness for upside options has coincided with a broad retrenchment in single stock implied (and realized) volatility.

Odd Lots of Charts

Last week I had the pleasure of appearing on the Odd Lots podcast with my former bosses/mentors Tracy Alloway and Joe Weisenthal to talk markets. If you’re reading this, you’re obviously a person who’s intellectually stimulated by markets. If that curiosity extends to the broader world of finance, economics, and business operations as well, I expect you’d get a kick out of listening.

Podcasts not being a medium that lend themselves well to visuals, I thought it might be useful to run through some charts pertaining to what we discussed. And on some topics (well, breadth) to avoid any repeats of things that have appeared here, there’s some fresh supplementary views as well:

Is brutal breadth a warning sign?

Higher rates (and oil prices, and food prices) are weighing a big mark on the stock market; just not the major indexes. It’s “bad momentum” — and a very high bar to clear — when all three of these are rising together, but the S&P 500 and Nasdaq 100 have demonstrated resilience in the face of rising yields.

If you had to choose between good breadth and bad indexes or bad breadth and good indexes, I think you’d have to prefer the latter.

“History demonstrates that there is essentially zero predictive power when these breadth statistics are close to current levels AND the market is in a bull trend (or when it is not),” writes Tallbacken Capital Advisors CEO Michael Purves. “We do not look at today’s distorted breadth as reason to be bearish on the SPX, but rather a reflection of a strong bull market underscored in part by substantial technology disruption.”

Goldman Sachs analysts led by Ben Snider also provide some fundamental footing for the degree of concentrated performance within tech that we’ve seen as of late. It’s the only sector where the median stock is seeing its expected profitability go up!

From a GS note on Friday:

“We expect most S&P 500 companies will report continued profit margin strength in Q3, but input cost pressures likely limited substantial sequential margin expansion. Consensus estimates show the median S&P 500 stock generating a net profit margin of 14.7% in Q3, compared with 15.1% during Q2. Despite contained wage growth, companies continue to face input cost pressures. Analysts have cut estimates for the median S&P 500 stock's Q3 profit margin by 11 bp since the start of the quarter, with estimates trimmed in every sector except Info Tech.”

Why else has megacap tech shaken off higher rates?

TL;DR: because the group got smashed during the first leg of rising rates, and there’s only so long you can keep a good bottom line down.

As of early September, the trailing three-month performance of the long-short US earnings revisions factor was -5%. Usually, the factor only does meaningfully worse than this when we’re emerging from a bear market or growth scare and investors are front-running a broad-based improvement in profits. So, in as much as there’s a “natural limit” to anything that happens in the market, you could argue that the selling of AI stocks with robust revisions had neared exhaustion levels.

The timing of the momentum break in the AI trade also coincided with a material trend wider for hyperscaler spreads, which have since largely steadied.

2026 was supposed to be the year of massive IPOs (and is still shaping up to be), but so far, the market has had to digest far more US IG supply from Mag 7 hyperscalers and Oracle (around $150 billion) than from the public debuts of SpaceX and Cerebras Systems (around $91 billion).

If you had to pick a debt-fueled source of profit fuel, the Mag 7 and developed market governments would probably be the safest options you could go for — and that’s what we’ve got right now.

The lag between capex spending and depreciation costs makes the AI boom an intense source of profit power that’s relatively self-contained within the industry (with some notable leakages to industrials and utilities).

Owning bonds has been all pain, no gain

I have a pretty consistent elevator pitch on the bond market: it’s a good news story about global growth.

Therefore, the rise in yields is more a symptom of the strong financial performance buoying stocks than a cause of potential downside.

(For Europe, however, Matt Klein makes the argument that some countries, notably France, are facing outsized pressure on long-term borrowing costs because of a worsening budgetary outlook. We’ll get back to that…)

At the risk of over-extrapolating recent performance, the case for a continued reset in yields to the upside might come from all the scars that bond buyers have picked up from stepping on rakes for so long.

Long-term bonds aren’t just competing with stocks for investor capital, they’re also competing with other bonds! And for shorter-term maturities, we’re basically looking at a lost decade.

The below chart shows what the “realized” real yield was on a five-year Treasury bond bought five years ago and held to maturity (that is, the August 2026 data shows the result for a bond purchased in August 2021).

From 2011 through 2021, inflation ended up destroying your coupon unless you bought five-year bonds right ahead of the shale oil bust.

European equal weight vs US

It’s no secret that I’ve been suspicious of the resilience in European stocks (particularly financials) given the concurrent rises in yields, oil as well as refined product, and agricultural commodities in Q3.

We’re seen a bit of a catch-down trade from European equal weight vs. the US as of late (with European financials lagging their US counterparts), but it still genuinely boggles the mind that European equal weight is closer to its closing peak than US equal weight.

When to run for the hills?

It’s uncontroversial to say that earnings drive stocks over the long term…but I am a simple person who contends that they’re also one of the better guides we have for short term performance as well!

A common question I field is, “what will you look at to know the AI boom is over?”

Loosely speaking, I have two answers: “whenever S&P 500 forward earnings estimates are down 5% or if there are credit events that suggest a boom fueled by debt is at risk of unraveling.”

The second half is more subjective, so let’s start with the track record on the first.

The last five times the S&P 500’s earnings estimates have been down 5% from their 52 week high, the benchmark index is also well off its 52 week high — typically, down much more than profit projections.

Earnings estimates are by no means a leading indicator. But if you’ve been invested in the market and enjoying the fruits of this profit explosion for years…do you really need a leading indicator?

In three of these instances, two year trailing returns even through that point have been positive. The best defense against downturns is having been there for the rally.

The pandemic is one of two exceptions. COVID is a relatively unique case based on the speed with which large swaths of the economy effectively shut down.

The other is the financial crisis, where the S&P 500 peaked in October 2007 while earnings estimates didn’t meaningfully roll over until almost a year later.

Things happened in between. And that’s where the more qualitative criterion comes into play…

New Century, the largest independent provider of subprime mortgages, declared bankruptcy in April 2007. Bear Stearns sold at a fire sale price less than a year later, while the leaders of Merrill and Citigroup both exited stage left in between.

You might think Oracle citing “force majeure” is a canary in the coal mine for our present time, but lots of other birds are singing a different tune.

Concerns these days center on how financing is being expanded (key word: expanded!) through so-called circular financing and other forms of credit wrapping; we also have execs leaving seemingly disrupted companies to go to the ones at the heart of the boom.

Whispers from Wall Street

From Jeff Jacobson, head of derivatives strategy at 22V Research:

“Here are some reasons why I believe the KRE could rebound from here. First, the recent pullback for the sector has taken KRE right to the November 2025 uptrend support level. We saw a similar sharp decline in Feb-March for the sector, only to see a subsequent 14%+ rally over the following month or so. Second, not only has KRE pulled back to what appears to be a support area, but the KRE/SPY relative spread also pulled back to just above the June lows. Back in June, we saw the relative spread spike higher as KRE rallied another 15% from the June lows to the July highs…
What stands out to me regarding KRE implied volatility is we have seen 2-month November volatility move higher, even as 2-month realized volatility has continued to move lower. This dynamic means you want to be a net seller of volatility in the context of adding upside exposure ahead of earnings season. Since the sector has had a sharp decline already, I favor selling a downside put to help fund an upside call spread purchase to help offset the elevated volatility…
Here is a November KRE trade I would consider at this time:
Sell KRE Nov 20th 66 put
Buy KRE Nov 20th 73/79 call spread
Costs ~ $0.50 (KRE 70.78 Fri close ref)”