1The ceasefire market went to zero, and it took the part of inflation that was cooling with it
Polymarket's US-Iran ceasefire ladder resolved against every leg overnight. The contract on the ceasefire holding through 31 August fell 98.7 points to zero. The 15 September leg fell 88.5 points to zero, the 30 September leg 77.5 points, and the 31 October leg 69.5 points. The event traded $1.58m in the last 24 hours against $6.31m of lifetime volume, so this was not a thin book moving on nothing. Yesterday the mid-September leg was priced at 90.5% and rising.
The sequence behind it is dated. Reuters reported Iran's Revolutionary Guards vowing a response to a US strike on Larak island on Sunday evening, then the Guards saying they had attacked two US bases in Jordan. The UAE said its forces dealt with a drone coming from Iran and called the attack a dangerous escalation. At 02:19 UTC the president posted "Kharg Island being blown to smithereens!!!" — Kharg is the terminal through which the bulk of Iran's crude exports leave.
Crude responded. Yahoo Finance's markets page showed the October contract up 2.77% at $85.71, and TradingKey reported Brent at $90. Reuters headlined its morning bid "Oil climbs as missiles fly in the Gulf", and carried a separate piece arguing oil holds above $80 while Middle East supply risk persists. rose more than 2% before the open. Europe's STOXX 600 fell on the oil move and still closed a fifth consecutive monthly gain.
Two contracts price how long this lasts. Hormuz traffic returning to normal by 15 September sits at 0.75% on $1.10m of volume, which is a market saying no. Traffic returning to normal by 31 December sits at 28.5%, down a point on the day, on $10.07m of volume and $361k of resting liquidity. A separate contract on a US invasion of Iran before 2027 rose three points to 15.5%, on $63.4m of lifetime volume and $775k of liquidity — the deepest book checked today.
The consequence runs straight into the rate question below. Energy supply was the component doing most of the recent cooling in the inflation prints. An escalation that puts Iran's main export terminal in play takes that back, and it does so two weeks before a meeting the market has just repriced.
2A September rate rise is now the favourite, and the long end went the other way
Polymarket's September Fed contract prices a 25bp increase at 58.5%, up eleven points on the day, against no change at 41.5%, down nine. Every cut leg together is under 1%. The increase leg was 43.5% on Friday and 47.5% on Saturday, so it has now crossed even money for the first time. The annual contract moved with it: a hike at some point in 2026 is at 74%, up eight and a half points on $8.27m of lifetime volume, and the no-hikes-in-2026 leg fell nine and a half points to 24.5%.
What moved it was Friday's Jackson Hole speech. Ross Givens, on his own channel, read it as a chair breaking with his own institution: Kevin Warsh placed responsibility for 65 months of elevated inflation on the Federal Reserve rather than on supply chains, the pandemic or Washington. Givens pulled out the numbers that make a cut hard to argue. The Fed's preferred inflation gauge is running 3.7% over twelve months and 4.1% over six, so it is accelerating rather than cooling. Of the 199 components in that basket, 54% are rising faster than 3% a year, against a pre-pandemic average of 30 to 32%. The effective federal funds rate is 3.63% against inflation of 3.7%, which makes the real policy rate negative — and the Fed's own series confirms the July federal funds print at 3.63%. Warsh's own phrase was that he would be "hard-pressed to describe financial conditions as restrictive". He cited business capital spending on equipment and software up around 9% over four quarters, the fastest since 2021, with more than half of it from the AI buildout.
Givens says implied hike odds moved from 35% to 46% within an hour of the speech. The prediction market has carried that another twelve points over the weekend.
The bond market's answer is the part worth arguing about. Givens describes the two-year yield rising to just under 4.30%, its highest in a month, while the thirty-year fell — a hawkish speech met with buying at the long end, which he reads as belief rather than fear. On that reading the market thinks Warsh will get inflation down and is willing to accept the growth cost. The Fed's own series supports the shape: the ten-year-minus-two-year spread narrowed from 0.47 to 0.39 on 28 August, eight basis points of flattening in one session, with the two-year at 4.20% and the thirty-year at 5.19% the day before.
Kevin Muir, of The MacroTourist, on Excess Returns, reads the same long end as unremarkable. Nominal GDP has gone from around four and a half per cent a year ago to six and a half now. The federal deficit has stayed elevated, corporate issuance is at a record on the back of the AI buildout, and the rest of the developed world has abandoned the two-and-a-half-per-cent deficits it used to run. Against that supply, he puts the thirty-year at roughly 5.30% and calls the move about fifty basis points: "I'm surprised it's not a lot worse." He adds that the gold market "got the memo real quick", reading political pressure on the Fed as a first step toward monetising the debt.
The two readings are not compatible. One says the long end is pricing a chair who will win. The other says it is pricing issuance and nominal growth, and has nothing to do with credibility. The 16 September decision, and what the thirty-year does in the week after it, is what separates them.
3The risk in the AI trade may sit in the earnings rather than the multiple
Muir, on the same show, made an accounting argument and credited Jim Chanos for it. When a hyperscaler borrows to build a data centre, the money it spends becomes revenue at the supplier immediately — he named and — while the buyer capitalises the asset and depreciates it over five, seven or ten years. The larger the buildout, the more aggregate earnings are flattered at the front, and the expense arrives later. Analysts then raise both the earnings estimate and the multiple, which counts the same optimism twice.
His summary is that this is not a price bubble. "The stocks are cheap in terms of the multiples" is the objection he keeps hearing, and his answer is that the denominator is the problem. He points at as the tell: its share count fell for ten straight years and is now rising for the first time, alongside aggressive borrowing, which he says is not what a cash-flow machine does. His extrapolation is that prices inflect lower before earnings do, and his precedent is the Gartner report that revised the internet's doubling time from 180 days to 270. The growth rate slowed. It did not reverse. That was enough.
Felix Breen, on Felix & Friends, reached the same caution from a different direction, quoting a Berkshire Hathaway director he did not name: the three obvious winners of the internet were Cisco, Yahoo and AOL; two were vaporised and the third never recovered; and the technology was real the whole time.
Dan Rasmussen, on Excess Returns, supplied the institutional version. He describes investment committees as "momentum investors with a three-year lag", and traces where that took private equity. Energy funds were the fashionable allocation in 2014 and 2015 and blew up in 2016, after which software returns at Thoma Bravo and Vista pulled the money in, until by 2021 more than 40% of private equity sat in software or healthcare software at high prices. His scepticism is about what was actually bought — not Salesforce or Adobe but, in his phrase, a company that "had cornered the market on car dealership software in the tri-state area". His line on the moat argument is the sharpest thing in the episode: "newspapers were recurring revenue".
There is a live counterexample. Bryce Leske and Alec Renehan, on Equity Mates, walked through 's June quarter: gross merchandise volume of $115.6bn, up 32% year on year, revenue of $3.58bn, up 34%, a fifth consecutive quarter of merchandise growth above 30%, and guidance above 30% against an expectation nearer 25%. Strip out equity investments and quarterly profit went from $568m to just over $1bn. Their read is that the fear of AI agents disintermediating commerce software has not shown up in the numbers, and Renehan called the wider scare "overblown". The caveat came in the same breath: the stock trades at roughly 100 times trailing earnings, which he flagged explicitly as a reason not to buy it today, even though he intends to keep holding it.
The two positions are testable against each other. If the earnings are the bubble, they hold up right until capex growth merely decelerates, and it is the decelerating quarter that matters rather than a falling one.
4A California bill that did not pass cost a utility a quarter of its value in two sessions
closed Friday at $16.62 and traded at $13.52 by early Monday afternoon, down 18.6% on the day and roughly 25% from Thursday's close of $17.97. The news aggregators carried the move with no driver attached. The driver is a legislative outcome.
California's legislature ended its session on 29 August by passing SB 492, which stopped well short of the wildfire-liability reform utilities had lobbied for. The proposal that would have restricted insurers from pursuing subrogation claims against electric utilities for wildfire losses was rejected, leaving the company facing insurer suits with no statutory cap. Wells Fargo cut the stock to Equal Weight, and it set a 52-week low at $13.31.
The mechanism is the opposite of the week's other stories. Nothing about the company's operations changed between Thursday and Monday. What changed is the size of a liability the market cannot bound, which is a repricing of a tail rather than of a business. It is also the only single-name move in this digest with a dated, documentary cause.
Wednesday is when the earnings argument gets a number
reports after Wednesday's close, and it is the cleanest read available this week on whether AI-driven revenue is still accelerating or merely large. and report after Tuesday's close, with and before it. , and join on Wednesday, and and report after Thursday's close. reported before Monday's open.
The rate argument waits longer. The decision that separates the two readings of the long end lands on 16 September. Between now and then the oil price does the work, because an escalation that reaches Iran's export terminal pushes on the one inflation component that had been falling.
