The Index Is at Its High, My Stocks Are Down: Odd Lots on Why Markets Move This Fast
On the October 1, 2026 Bloomberg Odd Lots, Sherwood News's Luke Kawa unpacks a market 0.5% from its high with only 51.2% of stocks above their 200-day, and half of this year's earnings growth tied to hyperscaler capex. Educational listening notes, no investment advice, no single-stock calls.
Contents
On the Luntai road the ninth-month wind howls all night; a riverbed of broken stones, each the size of a peck measure, goes rolling across the ground wherever the wind takes it.
—— Cen Shen, “Song of the Zouma River, Sending Off General Feng on His Western Campaign” (Tang, c. 8th century, translated by the author)
On the Bloomberg Odd Lots episode published October 1, 2026, hosts Joe Weisenthal and Tracy Alloway talk through the current market with Luke Kawa, head of markets at Sherwood News; the recording date was September 29. Kawa’s one word for this market is speed, and his numbers are these: heading into that week the S&P 500 sat 0.5% from a record high while just 51.2% of its members were above their 200-day moving average, and he cites a Goldman Sachs estimate that roughly half of this year’s earnings growth traces to hyperscaler capital spending. He also supplies the counterexample himself — breadth was this bad in 1998 and the market ran for another two years — so these figures tell you how few names are carrying the tape, not when the carrying stops.
I Was Reading “The Index Held” as “Everyone Held”
Going in, my picture of this year was simple. A war with Iran, the ten-year above 5%, an AI capability jump every few weeks — none of it dented the index, so the market had grown immune to rates and to geopolitics. I had written something close to that in my own notes: this market only prices AI, everything else is noise.
It sounded right because my control group was the index. The S&P is up more than 12% this year, and the news flow is thick enough that even a senior strategist changing firms barely registers — Joe mentions that David Zervos leaving Jefferies on September 28 would have been the biggest Wall Street story of the day ten years ago and this time went almost unremarked. Put those two together and “immune” is the conclusion you reach.
The flaw is that I used one average to describe the situation of several hundred companies. The index is a weighted average, a weighted average can be carried by a handful of names, and I had never looked at who was carrying it or how the rest were doing.
That 51.2%, and the Group Standing in the Corner
Kawa’s breadth work is where I noticed what I’d skipped. Half a percent from a high with 51.2% of stocks above their 200-day: the last time the market was that close to a high with that few names participating was the day after the dot-com peak. He says posting that annoyed a lot of people.
Then he hands over the rebuttal himself. Breadth was just as bad in 1998, and the market ran for two more years. That piece of reasoning was the most useful thing in the episode for me — if one indicator marks a top in 2000 and a midpoint in 1998, it can’t be your reason to leave. Kawa’s framing: breadth is neither a sufficient nor a necessary condition for a durable turn.
He also answers my “immune to rates” line directly. Parts of the market are not immune at all. The S&P equal weight isn’t. The Russell 2000 isn’t. Rates are a blunt tool, and they hurt a homebuilder far more than an AI company issuing mega-deals into the bond market. What I saw as immunity was cap weighting shrinking the companies in pain down to invisibility.
What that group looks like comes through in specifics. Joe pulls up McDonald’s, which peaked on March 3 and has gone straight down since, right around when the Iran war began (Tracy’s aside: wasn’t that when the CEO reluctantly ate part of a cheeseburger?). Nike was a $180 stock in 2021 and trades near $35. Kawa’s explanation is sharper than “the consumer is done”: these were relatively expensive stocks without exceptional top-line growth, in a world where nominal growth is accelerating. Why own a slow, expensive company when everything else is compounding fast? He adds a layer I hadn’t considered — Walmart and Costco spent a stretch serving as stand-ins for the software stocks AI was supposed to disrupt, because a fixed monthly spend at Walmart looks a lot like recurring revenue and Costco is a membership business. The market wanted to punish something, so it punished those. As software came back, their multiples deflated instead. His line: when AI is doing well, you have to find someone to punish.
Helene Meisler calls it the either/or market; Kawa calls it the Elaine-and-George market, where only one can be doing well at a time. There’s also a picture from the episode I keep coming back to: the index looks motionless while the feet churn below the surface, a duck. Across the third quarter, only two semiconductor companies in the S&P 500 made a 52-week high — AMD and Skyworks.
What I Watch Now, and Two Things I Can’t Settle
Late in the episode Tracy asks what would count as a real turn. Kawa’s answer is boring by his own description: earnings. He ran the test — if you waited until S&P 500 earnings revisions were 5% off their 52-week high, knowing analysts are late and knowing you’ve missed the top, how bad is that? Across the last six or seven bear markets, the only place you really get killed is COVID, because it happened too fast and the lockdowns were already underway. Otherwise, holding and looking back two years from that late arrival still shows healthy gains. I’ll use that, because it’s a bell that rings, while “breadth is terrible” is a mood.
The second thing I watch is the shape of the financing. Kawa describes a high-conviction call he made earlier: once hyperscaler free cash flow went negative, there were two paths — the AI boom slows, or the financing gets more creative. The financing got more creative. Then he lays out where profits are coming from: a large block from hyperscaler capex, increasingly debt financed, and another large block from the fiscal deficit. His question is blunt — can you think of safer places to be sourcing profits from than big-tech debt and government debt? Very safe, until it isn’t, which is the Minsky formula. With Oracle’s CDS spiking and NVIDIA working to credit-wrap the ecosystem’s borrowers, I’d treat the price of that debt as the thermometer for this cycle rather than counting new highs.
So if you’re like me, holding a basket that’s flat or down while the index prints highs, the question underneath is whether you picked wrong. The episode suggests a better question: which side of the dispersion trade are your holdings on? One side is the multi-layer AI beneficiaries — hardware, power, optics, and software has rejoined them. The other side is what Kawa calls inertia stocks: expensive, growing at an ordinary pace, with no link to the impulse. For a company on that second side, a bad chart is the market choosing between two uses of the same dollar. Goldman gives a ruler for that choice in a separate note: what explains multiple dispersion across the Russell 1000 right now is three-year-forward sales estimates, while one-year estimates and their revisions carry almost no explanatory power. Near-term growth is in the price; the bet being made is about three years out. Knowing which horizon your companies are priced on beats guessing daily why they fell.
The other common grievance is that bad news lands and nothing happens, so the market must be irrational. The Kawa–Alloway exchange gives this two sources at once. One is headline density — too many to digest, and a negative headline Monday is likely followed by a positive one Tuesday, so you ping-pong. The other is structure: retail participation, pod shops running tight stops, zero-to-four-day options, and the early agentic trading setups, which Kawa says look more momentum-driven than most. Stack those and you get more single-stock volatility, less index volatility, and mean reversion fast enough that you can miss it happening. His most honest line in the hour: the stock market is where we express views on the long-term earnings power of corporate America, and we do it with options that expire in four days or less.
Two things I can’t settle. First, inflation expectations. Yields jumped and US break-evens barely moved; both Joe and Luke say that surprises them. Kawa’s explanation is that the move came through real yields, and bonds need a higher real return to compete with equities for the same dollar. Reasonable — it still doesn’t explain why, after years of the Fed missing its target, the market extends that much credit to a long-run 2%. Second, the nature of this cycle. Housing used to be the business cycle, spilling into consumption through home equity and the wealth effect; Kawa says that whole mechanism has been transplanted onto AI, which is where the construction impulse and the wealth impulse now come from, and the share of American wealth held in stocks crossed above the share held in housing earlier this year. Tracy names what unsettles her: circularity plus size — stock gains fund spending, spending supports the economy, the economy supports stocks. Kawa’s closing thought stays with me: all of this has to be good for somebody downstream eventually, because there’s no profit without a consumer who wants the thing and has the income to buy it.
Worth Reading Alongside
- The Bloomberg Odd Lots episode of October 1, 2026, with guest Luke Kawa, recorded September 29. The show runs a daily newsletter covering each episode.
- The two Goldman Sachs notes referenced on air: this year’s earnings growth versus hyperscaler capex, and what explains Russell 1000 multiples.
- Kawa’s own market writing at Sherwood News on breadth and the dispersion trade.
- For the scale of the capex wave, Adam Tooze’s newsletter has compared it with past infrastructure booms; Joe also cites an economist’s estimate that a handful of companies would need revenue equal to about 9% of GDP to justify current spending plans.
One Thing to Take With You
One idea: the whole you belong to getting better does not mean you are getting better, and the whole getting worse does not mean you are getting worse. An average can be carried by a couple of components, and anyone reading only the average believes they have already seen everything.
Something I’ve tried: take one thing you recently waved off as “overall, fine” — your team’s quarter, a kid’s term, your own health, the mood at home — and break it into five to seven line items on paper. Mark each with a single direction: up, flat, down. Then look at which one or two items are holding up that “fine,” and how long the down ones have gone unmentioned. Last time I did this on my own health, the only item holding up “not bad” was my weight.
This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.