1The Treasury doubled its own long-bond buybacks, and whether that is plumbing or rescue is unsettled
On 19 August the US Treasury said it will at least double the size of its long-end buyback operations, from $2bn to at least $4bn apiece, beginning 9 September and running through 4 November. The increase covers the 10-to-20-year and 20-to-30-year sectors, and Treasury Secretary Scott Bessent told CNBC the figure "could be more than the 4 billion per issue". The context is a long end that had just backed up hard: the 30-year printed 5.31% on 17 August, a 19-year high, then 5.19% on 19 August and 5.23% on 20 August. The 10-year went from 4.72% to 4.65% over the same two sessions.
The mechanism is not complicated. A buyer of announced size and no price sensitivity arrives at the long end. Bond prices rise, yields fall, and the return on money parked in government paper falls with them.
Two of this weekend's videos read that as distress. Ross Givens argues the buyback, not the White House crypto meeting held the same afternoon, is what broke bitcoin out of a range it had held for months — the stablecoin rules on Monday and the SEC framework on Tuesday both landed without moving it, and the move came on Wednesday morning with the debt-management announcement. He also reports Bessent saying on television, of the long-dated paper, “Nobody wants them”; that quote is his rendering and we could not source it. Felix Prehn, on Felix & Friends, calls the same operation money printing and walks the plumbing: short bills issued, the Fed buying bills, the Treasury turning that cash into long-bond purchases. Both videos end in a sales funnel — a free live training in Prehn's case, a five-dollar annual membership in Givens' — which is context for how the framing is pitched rather than a reason to discard the operation itself.
Andy Constan, on Excess Returns, reads the same yields the opposite way. Rising real yields are what a hot economy produces, he argues, because an investor would rather own equity in a data centre with real earnings potential than a bond, and that preference lifts real rates directly. On top of that sits heavy supply, corporate and government both. He is dismissive of the idea that the long end can spiral: at a two-to-ten-year spread of 200 basis points he says he would "load the boat" on the positive carry, and there is deep demand for duration at a price. That spread stood at 0.50 points on 21 August, so his level is 150 basis points away.
The same fact supports both readings. What separates them is whether yields fall on the operation and stay down after it lapses on 4 November.
2A September rate rise is priced near one-in-three, into a long end the Treasury is buying down
Polymarket's "Fed Decision in September" contract, which resolves on 16 September, carries $50.2m of volume and $2.6m of liquidity — the deepest book checked today. No change trades at 66.5%, down two points on the day. A 25 basis point increase trades at 32.5%, up two. The two cut legs together are worth 1.7%. A week ago the hold was 68.5% and the hike 30.5%, so the drift has been steady and one-directional: the only live alternative to a hold is a hike, priced at roughly nineteen times the combined probability of a cut.
The data underneath is genuinely mixed. July payrolls fell by 23,000, to 158.858m from 158.881m in June, while the unemployment rate improved to 4.1% from 4.2% in June and 4.3% in May. Headline CPI printed 332.813 in July against 332.568 in June, a 0.07% monthly rise. The effective fed funds rate has sat at 3.63% since May.
Liz Ann Sonders, on Excess Returns, argues this cycle cannot be read in aggregate at all. Manufacturing went through a goods recession offset by services; services have since rolled over while manufacturing picked up. Those sectoral recessions and expansions run at different times, which is what produces the rapid-fire rotations in the market. She also notes wage growth is now running below the rate of inflation, so labour has no near-term route to reclaiming share: employee compensation has fallen from roughly 65% of GDP in the 1970s to about 54–55%, while corporate profits have gone from 5–6% to 11–12%.
So the government is leaning on the long end while the market prices a chance of tightening at the short end. If both hold, the curve flattens from an already-thin 0.50 points, and the first place the argument shows up is that spread.
3The Hormuz headlines escalated while the Hormuz prices de-escalated
The escalation is real and it is dated. Reuters reported today that the US Treasury will broaden the scope of its secondary sanctions on Iran, that Iran has threatened 45 tankers with fines and confiscation, and that fewer than 20 ships transited the Strait of Hormuz over the weekend. TotalEnergies' chief executive said the company is moving heavily discounted crude through the Strait profitably, and a separate Reuters piece argues the debate about crude volumes masks the real shortage, which is in refined fuels.
The prices went the other way. WTI fell 2.49% to $84.89. On Polymarket, "Strait of Hormuz traffic returns to normal by December 31" trades at 35.5% on $9.1m of volume, up four points on the day, after falling thirteen points over the previous week. The 15 September version of the same contract sits at 1.55% and is effectively ruled out. The ceasefire ladder softened at the front and firmed at the back: through 31 August 91.5%, down two points; 15 September 81.5%, down three; 30 September 74.5%, down one; 31 October 65.5%, up two.
That split is the useful part. The near-dated ceasefire legs slipped while the far-dated normalisation leg rose, which is a market pricing sanctions as escalation now and resolution later.
Bryce Leske and Alec Renehan, on Equity Mates, worked the other side of the same trade. A listener pitched Scorpio Tankers , and the numbers are striking: 74 tankers, a five-times price-to-earnings ratio, $1.3bn of net cash against $2.9bn of net debt in 2021, MR day rates of $52,270 against roughly $24,000 two quarters ago and LR2 rates of $77,749 against roughly $34,000, on a fleet-wide break-even near $11,000 a day. They declined to add it, on the grounds that a reopening deflates those rates and the asymmetry over five months runs against them. It went on a watch list instead.
The tanker equity and the Hormuz contract are the same bet with opposite signs. Both currently price a waterway that stays broken into the autumn and not into next year.
4In both AI and cancer vaccines, the argument is that the scarce input captures the economics
Bob Robotti, founder of Robotti & Company, told Excess Returns that the AI trade may not be AI. His case is that data applications are driving demand for physical things that have been underinvested in for ten or twenty years — cement, aluminium, copper — and that North America's structurally cheap natural gas is a durable industrial advantage, because gas cannot be moved as fast as it can be produced. His example is an aluminium plant in Brazil that cut capacity by half for want of gas supply, which matters because aluminium is a critical input to the renewables that energy insecurity is supposed to accelerate. Constan arrives at the same physical demand from the opposite direction, treating data-centre capex as one of the forces lifting real yields. Neither named a ticker.
The second case is more concrete. Moderna and Merck announced last Wednesday that their mRNA-based cancer vaccine, tested alongside Merck's Keytruda, cut the spread of melanoma; detailed results are due at a medical conference. Equity Mates put the share move at 177% in a day, against roughly 13.5% of free float held short and about $5bn of losses on those positions — and noted the stock is still down 60% from its high. A Google News item the same morning had the pre-market move at 89% with Merck up 8%, so the intraday and closing figures differ; the podcast's is the day's number. Renehan's read is that cheap candidate discovery pushes the choke point downstream into late-stage trials, which is why Moderna partnered with Merck rather than running the trial alone, and why he favours Eli Lilly — a stated preference in a portfolio discussion rather than a call to buy.
Both arguments have the same shape: the breakthrough happens at the top and the durable economics accrue to whoever owns the scarce capability underneath. The next two quarters of capex disclosure and trial data are what test it.
The buyback thesis meets its first live test before the Treasury buys a single bond
Nvidia reports on Wednesday 26 August after the close, on a $2.13 consensus, and Reuters was already naming it and Iran as the reason technology dragged the S&P 500 and Nasdaq lower today. Salesforce and HP report the same afternoon; Intuit and Zoom tomorrow after the close; Marvell, Autodesk, Ulta and Dollar General on Thursday. None of that touches the bond argument, and neither does the calendar: the enlarged buybacks do not begin until 9 September and the Fed does not decide until 16 September, so the next three weeks price expectations rather than events. Two dated items sit outside that window — gold, at $4,698.50 and still climbing, is running into US inflation data, and Trump said today that tariffs on all Canadian cars, trucks and automotive parts rise to 50% on 1 January 2027.
