Risk and Return Sep 18, 2026 44m 25m saved
Traders short the middle of the Treasury curve are up about half a billion dollars, on S3 Data's reading of shares shorted across the four main Treasury exchange-traded funds, a series it indexes to 12 June 2026.
Shipping is the mirror image of that trade. Long positions have piled in on the view that the Iran conflict keeps freight rates elevated, and short sellers spent most of the year standing aside. September broke the pattern.
"A lot of people making a lot of money, they're up about half a billion bucks, and it's telling you what people think about what's going to happen here with interest rates."
Every positioning call in this episode rests on S3 Data, which tracks long and short positions across the market daily and, by the show's own description, distributes through Bloomberg and FactSet. That matters for how much weight to put on the September turn in shipping shorts: a daily series shows a change within the month, while the exchange short-interest filings most commentary runs on arrive twice a month and late.
The full episode is covered here so you can skip it. 44 minutes of audio, 19 minutes of reading.
Here are the 14 takeaways that matter.
Key Takeaways
Shares shorted across SHY, IEI, IEF and TLT show a flat short end and a heavily shorted belly
Those belly shorts are up about half a billion dollars and called the steepening right
The 10-year moved from above 5% to just below it on a rate increase at the short end
Shipping longs are heavily invested on the conflict, with shorts adding in September for the first time this year
Shipping short interest as a percentage of float has recovered to near its high
Oil's price and short interest were almost the same graph until the correlation broke this month
With oil above $100, the trade turned supply-driven rather than conflict-driven
The economy needs 3% growth for 10 years to work out of a $40 trillion hole, against forecasts of 2% to 2.5%
AI is on track for $100 billion faster than any new industry, which is the case for reading the safety warnings as positioning
A planned insider sale like Larry Ellison's is filed months or years ahead, so the timing carries no signal
1. The Crowded Shipping Long
Shipping stocks came first because they double as a reading on the economy the conflict has produced. The S3 series behind the chart is built daily rather than from periodic filings.
The data is a daily long and short series
And we track both long and short positions throughout the market on a daily basis.
A host
The template for what is happening in shipping came from earlier in the year, when active long positions in South Korea were gaining pace faster than short positions in Hong Kong. A gap opened between the two. Then longs fell, Hong Kong shorts rose, and the spread closed.
Freight rates have since gone the other way, and the long side has taken the obvious position.
The long case is that nothing resolves soon
The same thing's happening we think here, which is shipping rates are going through the moon.
A host
The reasoning attributed to those long positions was spelled out on air: the conflict does not look resolvable in the short term, so rates stay high, so pile into the sector.
What the longs are assuming
Look, the conflict here doesn't seem to be resolvable in the short term. That means shipping rates are going to be high.
A host
2. Shorts Rebuild In September
For most of the year the short side did not fight it. Positions were not cut to nothing, but they were not added to either. September is where the series changes.
The shorts stopped standing aside
And the short positions have more or less, I don't want to say given up the ghost, but they've certainly not increased their positions, except recently, which is interesting.
A host
Short interest as a percentage of float tells the same story from the other direction: it has recovered to somewhere near its high.
Short interest as a share of float is back near the top
On the other side, right, short interest as a percentage of float, we see that more or less, you've seen short positions recover somewhat near their high.
A host
So the position is two-sided, and the read offered was that the bears are timing something specific rather than arguing with the freight rates themselves.
Longs still piling in, bears picking up
So what is this all saying? It's saying people still bullish, longs still piling in, but a pickup in bearish sentiment on the short side.
A host
The bears may be positioning for after the midterms
So what is that all saying? Still really bullish on shipping but bearish sentiment sensing maybe after the midterms this talk is real.
A host
For anyone long the sector, the useful part is that the trade is now crowded on one side with a growing set of counterparties who expect the conflict premium to come out of freight rates at a date on the calendar.
3. The Oil Reversal Breaks
Oil has spent the year as a textbook reversal trade: long on the way up, short on the way down, driven by whatever came out of the administration about deals being on or off.
Price and short interest were the same chart
And you can see price and short interest here is just almost locked up until now.
A host
The correlation was tight enough that the two lines were hard to separate.
Two lines, one graph
You have price and shortages very highly correlated. They're almost the same graph.
A host
Traders were not forecasting oil. They were trading headlines about peace plans, then the absence of peace plans, then the conflict degrading and escalating again, with the short book flipping each time.
4. Supply Takes Over From War
Above $100 the relationship comes apart. Price keeps rising while short interest falls, which is the signature of covering rather than of a new directional view.
The trade is no longer about the conflict
We think this is very telling because it says that the trade is no longer purely about the conflict. It's actually about the supply of oil. And that's a big shift.
A host
Two things are pushing in the same direction: reserves at what was described on air as an all-time low, and short sellers buying back stock into a rising price.
Covering on top of the supply problem
People are covering their short positions causing price to go higher on top of the supply concerns.
A host
One of the hosts said his newsroom's reporting had JP Morgan unable to work out where the price is heading. The answer from the data side was that this is exactly what the broken correlation implies. A conflict-driven oil market could at least be traded off the news flow; a supply-driven one removes the signal that made the reversal pattern work.
Harder to gauge, not easier
Now it's about supply. So it makes it even harder to gauge where price is going to go.
A host
The instruction to listeners was narrow and repeatable: watch price against short interest, because the gap between them is the tell.
What to watch from here
Keep an eye on price and relation of short interest.
A host
5. Four Treasury ETFs, One Bet
The rates chart is built from changes in short positions across four funds that between them cover the curve.
The four funds and what each covers
You have the SHY which is the 1 to 3-year, IEI which is 3 to 7, IEF is 7 to 10, the TLT which is the long end of the curve
A host
Shares shorted in each are indexed to 12 June 2026, which makes the shape of the positioning visible rather than the absolute size. The short end is flat. The belly, where duration sensitivity concentrates, is where the shorts went on in size.
The short end flat, the belly loaded
basically the short end of the curve is flat and then the belly of the curve, where people think there's a lot of interest rate sensitivity, shorts increased quite a bit
A host
Where the sensitivity sits
The middle of the curve is where the most sensitivity is
A host
6. The Belly Was Dead Right
The position worked. The mark quoted on air was about half a billion dollars of profit, and the point of quoting it was that the money was a forecast as much as a return.
Half a billion dollars, and dead right
A lot of people making a lot of money, they're up about half a billion bucks, and it's telling you what people think about what's going to happen here with interest rates.
A host
What the position was expressing was a steepening curve, which is a view that had been argued for a long time before it paid.
The call inside the position
Oh, that tells you there's a curve steepening going to happen and they've been saying that for a long time and it happens.
A host
The broader claim is about where to look for that kind of signal. The bond market said one thing; the ETF hedging flow said the same thing, in a series available daily.
Two sources, one message
So pretty interesting that not only were the bond markets telling you this, but the way people use ETFs to hedge were also telling you that too.
A host
7. A Hike The 10-Year Liked
The long end did not behave the way a rate increase is supposed to make it behave. The 10-year had been above 5%. After the hike at the short end it sat just below.
The reaction that argues the hike worked
It was above 5% as you know, now it's just below 5%, on a rate increase on the short end, which makes me think that Warsh did the right thing.
A host
The move down was small.
How small the move was
It actually came down like triflingly.
A host
The interpretation is that the hike bought credibility on inflation, and that credibility is what keeps the long end contained. The counterfactual was offered as a guess rather than a claim, which is the honest version of it.
The counterfactual nobody can settle
You don't know, if he didn't raise rates maybe the 10-year would have gapped out to 5.5%.
A host
A related complaint ran underneath all of this: the political focus stays on the policy rate while consumer borrowing is priced off the 10-year. The anecdote used to make the point was Treasury Secretary Scott Bessent telling an economic club in New York, two years or so ago, that he had finally got the president to stop obsessing about the Fed funds rate and worry about the long end instead.
What actually prices consumer credit
Trump is still obsessed about the Fed funds rate when it's the 10-year that all this consumer stuff is pegged off
A host
8. The $40 Trillion Hole
The growth arithmetic is the reason the inflation path matters. Earnings have been good, and the optimistic case is 3% growth. Most forecasts are lower.
The growth rate the arithmetic needs
So, we need to grow at 3% for the next 10 years to work ourselves out of a $40 trillion hole.
A host
Forecasts were put at two to two and a half percent, and the odds of hitting 3% for a decade were rated honestly.
Probably not, was the answer
Now, is that going to happen? Probably not.
A host
The fallback is that halving the number, or getting it to something like $15 trillion, would still count as a good outcome. Which is why the second half of the requirement carries as much weight as the first.
Growth is only half of it
So, we need growth, but we also need low inflation, right? That's the magic number.
A host
9. From 3.7% To 2%
The argument for why the long end is behaving is that the new chair identified the one variable bond investors care about and committed to it in public.
What the bond market was told
Working in tandem, what Warsh has correctly identified is what's on the minds of bond investors, inflation.
A host
The number and the mandate
But we're going to get it down from 3.7% to 2% or close to 2% and that's my mandate and I'll do that. And bond markets love that.
A host
If that holds, the consequences run straight into housing and mortgage rates rather than staying in the rates market.
The best of all possible worlds, conditionally
So you might have the best of all possible worlds here, which is that the long end is under control.
A host
What a contained long end protects
It doesn't blow out. It doesn't crush the housing starts or sale of homes or mortgages.
A host
10. Gas Is The Third Leg
Two legs of the stool are growth and interest-rate policy on inflation. The third is the pump price, and it is the one that neither the Fed nor earnings can fix.
The last leg
And the last piece in the puzzle is fixed gas prices.
A host
Where the pump price is set
The cost of gasoline is highly correlated to what's going on in Iran.
A host
The tax nobody voted for
And maybe that's just a tax we have to live with because we want to see a stabilized Middle East.
A host
Against that, one of the hosts relayed what he had heard from people close to the national security apparatus: Iran has been degraded far enough that its capacity to finance activity abroad is gone for a long time, even if the oil price does not reflect it yet.
Degraded, whatever the oil price says
we have so militarily degraded Iran that it may not show now in oil prices
A host
What is left
It doesn't have the resources. It doesn't have the army. It doesn't have the money.
A host
11. Talking Their Own Book
The AI safety statements that moved markets this week got the same treatment as any other positioning. The economics come first.
The industry's economics explain the volume
Tens of billions of dollars on track to make a hundred billion dollars faster than anything that we've seen as a new industry.
A host
Revenue growth with no precedent
You have maybe the fastest growth in revenue of any industry ever, ever.
A host
From there the behavior follows: an industry that large, growing that fast, with competitors split on open versus closed and on frontier models against China, produces public statements that are competitive moves.
The obvious reading
So, it makes total logical sense that the main players are jockeying for a position, looking to talk their own book, understanding their own competitive weaknesses.
A host
The specific mechanism named was asymmetry. A company whose costs are rising while the price of a token falls has more reason to want the rules rewritten than a company winning on cost.
Who has the incentive to shout
And I think the weaker players, at least this is my read, have figured out the largest megaphone.
A host
The counter-argument on the same side of the table was that stopping is not an available option, because the failure modes that prompted the warnings are themselves engineering problems.
The fix is forward, not backward
the only way to solve something like that is to advance the technology.
A host
One practical note for anyone holding AI exposure across the private names: single listings do not stay single. The example given was SpaceX, and the expectation was the same pattern on an OpenAI listing.
Why one IPO reprices the whole group
And any one of the IPOs affects everybody. We saw that with SpaceX.
A host
12. A Very Rehearsed Rollout
The part that raised suspicion was not the content of the warnings. It was the production quality.
Too good to be spontaneous
Very well rehearsed, very, very, very, very well communicated.
A host
Two details were offered in support. The tweets were with the Wall Street Journal sixteen minutes before the statements went live, and the people delivering them performed like anchors rather than like researchers.
The timing of the press placement
the Wall Street Journal had the tweets 16 minutes before this thing went live
A host
Why that is the tell
Now, we know a lot of data scientists at S3, okay, I love them, but the last thing I would do is put them on national TV.
A host
Because it is genuinely hard
No one does it flawlessly without training and practice.
A host
The objection to the whole exercise was framed as a cost rather than as a disagreement about risk, and the closing example was a risk-taker who lost roughly $300 million in 1987 and went on to build the largest asset manager in the world.
The price of being scared off risk
By the way, if Larry Fink gave up risk-taking after he lost like $300 million in '87, we wouldn't have BlackRock.
A host
A second objection was procedural. If the companies cannot be trusted to set their own guardrails, the alternative is an international agreement, and the participant who matters is not signed up to the last one.
Why a global deal is not the answer
Well, okay, that means there's got to be a global deal. But that's not going to happen for five years.
A host
The precedent on multilateral deals
They're not even a member of the climate accords.
A host
13. The Ellison Filing
The stock sale that unsettled Silicon Valley was a planned one. About $7 billion of Oracle stock, set in motion in May, with the filing reaching the market months later, and then a public reversal.
The mechanics nobody looked up
These things are done months and months, if not years, in advance.
A host
What a filing is not
Like he didn't wake up two weeks ago and go to sell.
A host
The point is that the timing of a planned sale carries no information, because the one thing an insider cannot do is sell at short notice.
The rule the timing follows from
So, the last thing you want to do as an insider is file to sell your stock and do so suddenly.
A host
The estate-planning explanation was straightforward, and one of the uses named for the money was funding his son's takeover of Warner Bros. Discovery. What made the episode notable was the reaction: a fortune around $200 billion, roughly half what it was a year ago after the correction in Oracle stock, and a flip-flop read as a loss of nerve.
The reaction, not the sale
It got a lot of people in Silicon Valley worried that maybe Larry Ellison has lost his fastball here.
A host
14. October 1 In California
The Warner Bros. Discovery deal has a hard date on it, and the California attorney general's antitrust suit is the obstacle. The bid is being pressed at the state's Democratic establishment rather than in court, according to a New York Post report, with Governor Gavin Newsom, Xavier Becerra as the man widely expected to succeed him, and Los Angeles mayor Karen Bass all in the frame. The mayor has come out on the deal's side, which one of the hosts read as a planted move to get the attorney general off his refusal to negotiate anything short of structural change.
The date the deal dies
October 1st, which is the deadline that Ellison has set, if you don't get into serious discussions, we're out, he's leaving.
A host
The argument against the suit is that the combination is not plausibly a monopoly given where entertainment now sits against big technology and Netflix, and that regulators elsewhere, including some of the most interventionist in Europe, have said so.
The view of the case
I think the antitrust suit is stupid.
A host
Who can still stop it
But a bunch of attorney generals, like 12 attorney generals, including the guy from California, can stop this thing.
A host
The stake is employment. The number put on a Paramount exit from California was 50,000 jobs, and the conclusion drawn from the fight was about where production gets done next.
The jobs at risk
if Paramount leaves, they lose like 50,000 jobs
A host
The conclusion the fight invites
There's no reason to be in California.
A host
Bonus Insights
The frame used for the AI safety push was that a warning and an interest are not the same thing. The comparison was the physicists who understood what splitting the atom meant, were frightened of it, and wrote privately to Roosevelt rather than staging a media rollout. Against that, the objection to the current warnings was cultural: the national anthem calls this the home of the brave, and a country in that position does not shut the future down or announce that everyone is about to die. The medical case was offered as the concrete counterweight, on the grounds that AI-guided detection is already changing treatment in prostate and breast cancer, including avoiding invasive surgery. One of the hosts expected the administration to land nearer what Texas is doing on data centers, which he described as a rule that a project cannot take a rural community's water and electricity and has to self-generate instead, rather than anything that slows model development.
The rhetorical device used for the geopolitical section was Churchill on 10 May 1940: what is our aim, and what is our policy. The complaint was that the administration has stated neither on Iran, and that refusing to state them does not avoid being boxed in, it just means being boxed in by someone else. The speculation on timing was whether it is better to finish the conflict now or muddle through until after the midterms, with a working assumption of House seats lost and the Senate possibly held.
S3 had nothing to announce, which was itself a data point about how the firm operates: "Nothing to report here, but when there's something to say, we will say it." The one forward-looking line was about reach rather than strategy, with the business itself unchanged, new markets, newer products and greater reach.
The episode also spent several minutes on the reopening of San Pedro, a New York restaurant run by two Italian immigrant brothers and described as the cafeteria of Wall Street for decades, now rebuilding its customer base. It is not a market call, but it is the sort of item that tells you which rooms the show's reporting comes out of.
The bottom line is a set of positioning reads that point in one direction: the crowded trades are long shipping on a conflict premium and short the belly of the curve on a steepener, and only one of those has already paid. Oil has stopped being a news trade and become a supply trade, which makes it harder rather than easier to forecast. The rates view rests on a single condition, which is that the new chair delivers enough of the move from 3.7% to 2% to keep the long end quiet while the pump price does what the Middle East tells it to.
Products, Companies & Tools Mentioned
S3 Data (The daily long and short positioning series behind every chart in the episode; the show says it distributes through Bloomberg and FactSet)
SHY, IEI, IEF and TLT (The four Treasury ETFs in the curve chart — 1 to 3 year, 3 to 7, 7 to 10 and the long end — with shares shorted indexed to 12 June 2026)
Oracle (The stock behind the roughly $7 billion planned sale, and down enough over the year to halve the holder's fortune)
Warner Bros. Discovery and Paramount (The merger the California attorney general is suing to stop, with an October 1 walk-away date)
Netflix (Named as the reason the combination is not plausibly a monopoly, on the argument that it is big technology adjacent)
OpenAI and Anthropic (The two firms whose safety statements were read as competitive positioning; a listing by either is expected to reprice the group)
Hugging Face (The platform in the bot incident that prompted the warnings, described on air as an alleged catastrophe rather than a real one)
SpaceX (The precedent for one private listing repricing everything around it)
BlackRock (The closing argument against being scared off risk: a $300 million loss in 1987 did not stop it being built)
Bloomberg and FactSet (The two terminals the positioning data is distributed through)
New York Post (Where the reporting on the lobbying campaign around the California suit appeared)
JPMorgan (Reported as unable to work out where the oil price is heading, which the data side treated as confirmation of the broken correlation)
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: