1The September decision is a coin flip on a hike, and the rates market has repriced where equity volatility has not
Polymarket's September contract is the deepest book checked today — $65.2m of lifetime volume against $3.63m of resting liquidity. It prices a 25bp increase at 43.5%, up 13 points on the week, and no change at 54.5%, down 14. Every cut leg together is under 1%. A week ago this was a hold with an outside chance of a rise. The argument has moved: it is now about whether the Fed hikes, not about when it cuts.
Joseph Wang, on Fed Guy, reads Kevin Warsh's Jackson Hole speech as the cause and thinks the market still under-prices it. His own number for September is 90% against the market's 43.5%. He builds it from the text rather than the tone, and the three pieces are checkable. Warsh recommitted to 2% PCE as a fixed target, which closes off the goalpost-move traders had feared since his confirmation talk of trimmed measures. He described financial conditions as not restrictive, citing credit spreads and loan growth — on Wang's read, a chair saying policy is not biting is a chair saying he has not done enough. And he added a speed test: missing the target for 65 months does not count as returning to it in a timely fashion. Warsh also pre-empted the obvious objection, saying stable inflation expectations "can change very quickly" and are not grounds for comfort.
The rates tape gives partial support. The two-year yield rose 11 basis points on the day of the speech, on Wang's count. The 10-year-minus-2-year spread compressed from 0.47 to 0.39 on 28 August, an eight-basis-point flattening in one session, which is the front end repricing faster than the long end. The daily two-year and ten-year prints for the 28th have not published yet, so which leg did the work is not confirmable from the source data. The spread move is.
The split worth keeping is about who has already repriced. Wang argues the rates desks and the gold market understood the speech — gold fell hard on Friday — while equity investors have not, leaving "more repricing to go". CNBC reported the same afternoon that stock traders were warming to Warsh as the volatility index touched a year-to-date low. Those are two readings of one tape. Either equities have priced the hike and are relaxed about it, or they have not looked. Nothing in this window separates them; the 16 September decision does, and the labour data before it. Note also that Kalshi has been rate-limited for a seventh straight day, so every probability above comes from a single venue.
2Sixty-five billion barrels of reserves do not loosen a market that is short barrels this quarter
The largest supply announcement in years landed on Friday evening. Trump said the United States had secured majority control of more than 65 billion barrels of proven Venezuelan reserves at no cost to the taxpayer, framed it as more than doubling American reserves, and said it would "substantially lower gas prices". A White House official told CNN the structure is 55% of effective output from a new private joint venture, built on 100-year concessions granted to a private company that has not been named. He posted the claim twice inside this window.
Set against that, the barrels that clear this quarter are still constrained. Reuters reported the Navy blockade slashing Iranian oil exports as the administration shifts to economic warfare, Iran's navy claiming "full control" of the Strait of Hormuz, and QatarEnergy cancelling gas deliveries to Italy's Edison until early November over the war. Polymarket prices Hormuz traffic returning to normal by 31 December at 31.5% on $9.9m of volume and $434k of liquidity — up 3 points on the week, so the market got marginally less pessimistic, not more.
Which force dominates near-term is not close, and the reason is a distinction the announcement blurs. Reserves are not production. Venezuela's development runs on a permitting-and-drilling clock measured in years, and the concession length in the structure says so plainly. Hormuz is priced daily. So the near-term supply picture stays the war's, and the Venezuela agreement is a term-structure story about the back of the curve rather than a Q4 one. The political clock explains the timing: Reuters reported on 27 August that Trump was meeting refiners and fuel retailers because the war was pushing pump prices ahead of the midterms.
There is a clean market-versus-expert split here too. Wang argues the president holds a tail-risk instrument that can lift markets at any moment — declaring some understanding with Iran, which would take oil and rates down and equities up. Polymarket prices a US–Iran final nuclear deal by 30 September at 1.35%, on $16.8m of lifetime volume. The market is not saying the instrument does not exist. It is saying the odds of it firing inside a month are roughly one in seventy.
3The case against the AI trade is an argument about accounting, not about multiples
Kevin Muir, who writes The MacroTourist, put the bear case on Excess Returns in a form that does not depend on valuation at all, crediting the framing to Jim Chanos. When a hyperscaler borrows to build a data centre, the suppliers book the revenue immediately — chips, power equipment, construction — while the buyer depreciates the cost over five to ten years. So the bigger the buildout, the more reported earnings are flattered at both ends, and analysts raise their forward numbers into the flattering. Muir's conclusion is that the bubble sits in the earnings rather than the multiple, which means "the stocks are cheap on forward earnings" is not a defence when the denominator is the thing in question.
His supporting evidence is a balance-sheet change rather than a price. 's share count has fallen for a decade and is now rising for the first time, alongside aggressive borrowing — his line is that these are no longer the cash-flow machines people bought them as. is his worked example of the borrow-and-build pattern, with and named as immediate beneficiaries on the revenue side.
The forward-looking part is the useful part. Muir argues prices can fall before earnings do, and reaches for 2022: next-twelve-month EPS kept rising through the first months of that drawdown, and the market stopped falling only once earnings actually rolled over. If that sequence repeats, waiting for an earnings miss to confirm the thesis means waiting past the move.
What he is not doing matters as much as what he says. He named no short. He expressed the view in precious metals instead, stepping back into gold a couple of weeks ago on a fifty-day moving-average break after sitting out the late-2025 melt-up, and buying platinum on a below-cost-of-production argument with hybrid vehicles supporting catalytic-converter demand. His stated reason for owning gold is the People's Bank of China as a structural buyer, not real rates or the dollar. So the earnings-bubble call and the positioning are in different asset classes — worth noting before treating the argument as a trade.
Two counterweights sit in the same week's news. CNBC's read of this earnings round was that "the AI buildout is not a zero-sum game", and Jim Cramer said he was warming back to the AI trade without being ready to act. Muir's second warning is structural rather than fundamental: levered ETFs rebalance daily, so they buy into strength and sell into weakness, and he uses as the illustration of what a gap-down morning would force. He cites a two-times levered product that fell 45% and then 55% as the provider flattened it, and a Hong Kong three-times product that grew from $2bn to $30bn. His framing is that index concentration in one theme is the highest since 1929 and levered products are being layered on top of it. That is a claim about market plumbing; nothing this week tests it.
4The tariff fight with Canada runs against the trade data underneath it
Muir pulled the 2025 US Census Bureau figures rather than argue from headlines, and the numbers cut against the stated rationale. Across every automotive category combined, Canada runs a $7.2bn deficit — meaning the United States runs an auto surplus with Canada, not the reverse. The overall bilateral deficit is dominated by a single line of roughly $82bn in oil and gas, with the remaining categories in the fives, tens and twenties. Strip out oil and gas, which is a globally traded commodity with the refining value-add on the American side, and the deficit becomes a surplus in finished goods.
His policy point is separate from the arithmetic and is about volatility rather than level. Tariffs turned on, off, renegotiated and turned on again destroy the certainty a plant investment needs, so the on-off cadence works against the stated goal of returning manufacturing jobs. His own preference, stated plainly, is a flat 10% left alone.
The caveat travels with the claim. Muir is Canadian and says so, and he opens by conceding that Canada under-spent on defence for years and that the goal of rebuilding middle-American manufacturing is a fair one. The census figures are checkable; the framing around them is his.
The political read is where this connects to the rest of the day. Polymarket prices the Democrats to win the House at 88.5% on $9.9m of volume, and the Senate at roughly even — both parties quoted near 50.5% on a $4.0m book, which is a coin flip inside the spread. Wang's argument is that Michigan and Maine are Senate battlegrounds that trade heavily with Canada, which is why he expects a pre-midterm de-escalation with Canada or Iran rather than an escalation. Pointing the other way, the president spent two posts in this window attacking New York's governor for siding with Canada, "which has been ripping us off for decades on Trade and Tariffs".
The week ahead marks the hike to market before it marks the buildout
Markets are closed for the weekend, and the next session is Monday 31 August, which carries nothing of scale. The tests arrive after it. reports on 1 September against a $45.9bn revenue estimate, and on 2 September against $29.9bn — the two largest prints of the week and both direct reads on whether AI infrastructure demand is still accelerating or merely large. Neither settles the accounting argument in section 3, because that argument is precisely that strong supplier revenue is what an inflating earnings base looks like.
The Fed claim gets marked sooner and harder. The September meeting is 16 September, and the labour data before it is what moves a 43.5% probability. Note the tension already in the printed numbers: unemployment fell to 4.1% in July from 4.2%, which is the reading Warsh cited as the employment mandate being met, while payrolls over the same month fell by 23,000. A falling unemployment rate on falling payrolls is a shrinking labour force, and which of those two the next print emphasises is the thing to watch.
