1A September hike went from argument to near-lock in a single day
Polymarket's September FOMC contract prices a 25bp increase at 80.5%, up 16 points over twenty-four hours on $4.9m of daily volume against $1.27m of resting liquidity. The no-change leg fell almost exactly the mirror image, 17 points, to 18.5%. A cut is priced at 0.4%. A separate and much thinner contract on any hike during 2026 sits at 89%, up 11.5 points, and the market on no cuts at all in 2026 is at 93.25%.
The move has a dated cause. August CPI, released Friday morning, rose 0.40% on the month and 3.35% on the year — in line on the headline, which Polymarket's own resolved contract confirms at 3.4%, but core rose 0.29% against a 0.2% forecast. Thursday's producer prices landed in line, removing the last argument for waiting.
Two venues agree, which is the strongest thing about this reading. Brent Kochuba of SpotGamma, speaking on Excess Returns on Thursday afternoon, watched CME's FedWatch travel "from about 65% to north of 80%" on that morning's PPI print — an independent venue landing within a point of where Polymarket closed Friday. Kochuba's own gloss was that the increase itself is almost beside the point: a 25bp move is "sort of ceremonial" next to what is happening in yields and debt.
The front end did the work. Two-year yields rose 13bp on Thursday to 4.56%, against 12bp on the ten-year to 4.95% and 9bp on the thirty-year to 5.37%; by Friday the two-to-ten spread had compressed from 0.39 to 0.33. A curve that flattens while the whole curve rises is a market pricing a central bank into action rather than pricing a growth scare.
Katie Martin, markets columnist at the Financial Times, speaking on Prof G Markets, expects the increase but would not pin it to this meeting: "whether it comes in September I don't know, but I think there will be something before the end of the year." Her condition is the one that just resolved — "if inflation stays sticky and payrolls keep doing what they're doing, I don't see how Walsh has a choice."
So what: the argument has moved on from whether the Fed moves to whether one increase is the end of it. The October contract already says mostly yes — no change at 61.5%, a further increase at 36.5%.
2Friday's bond move was about real rates, not inflation expectations
Two of the week's sources disagree about what is actually weakening bonds, and Friday's tape splits the difference in a way that is worth recording.
Jens Nordvig, president of Vanda and founder of Exante Data, on The Compound, pointed to compensation for inflation as a live and rising factor: five-year breakevens "was 2.2, now it's 2.4, 2.45", and he tied that to the Fed's problem — "the AI inflation is here, the energy inflation is there, and they're coming from an inflation level that is already above target."
Katie Martin, on Prof G Markets, reported the opposite from her own conversations: "inflation expectations have been pretty well contained. So it's not inflation that's been doing the heavy lifting in weakening bonds over the past few weeks." Her candidates are fiscal — borrowing volumes — plus what she called mixed messaging, with Fed chair Kevin Warsh welcoming higher yields in his first press conferences while Treasury Secretary Scott Bessent buys bonds back to push them down. "Someone tell me what to believe."
The prints back Martin for the day and Nordvig for the trend. Five-year breakevens did climb to 2.46% by Thursday, vindicating Nordvig's direction. Then on Friday, with the hot core print in hand, they fell to 2.40% — giving back the entire week's gain on the day inflation data landed. Meanwhile the ten-year real yield reached 2.55% on Thursday from 2.43% on Monday. Michael Batnick, on the same Compound episode, put that in its longer context: real rates are "the highest they've been since 2007."
So what: if compensation for inflation falls while nominal yields hold, the thing repricing is the real cost of money, not fear of prices. What would settle the disagreement is the next two breakeven prints: if they keep falling as the increase is delivered, the market is saying the Fed is credible and the pressure is fiscal, which is Martin's case.
3The oil market marked Iran down on the day the rates market marked the Fed up
The four videos were recorded Thursday, and the Iran picture they describe is one of escalation. Kochuba called it "flaring". Finnhub's wire carried Reuters reporting Houthi forces reaching a key island with a pipeline shut. Nordvig noted the conflict is now past six months against an expectation of weeks, and that the administration's position had been that once the Strait of Hormuz reopens, "WTI will crumble from 100 back to 70."
The prediction markets moved the other way on Friday, and they moved together. September contracts on WTI touching $105 fell 36 points to 53%; $110 fell 21.5 points to 34.5%; $120 fell 12 points to 12%. The mirrored downside contracts rose in step — WTI trading down to $95 this month gained 18 points to 81.5%, and down to $90 gained 18 points to 57.6%. Every rung of the Iran-Oman Hormuz management agreement ladder gained roughly 14 points on the day: a deal by 30 September now prices at 40% on $42k of daily volume, by 31 October at 53.5%. These oil contracts are small — $14k to $48k of resting liquidity each — so no single one carries much weight; eight of them moving the same way on the same afternoon does.
Polymarket's daily settlement ladder puts WTI's Friday close between $99 and $100, and the market on WTI touching $105 during the week of 7 September resolved no. That resolves a conflict in the day's headlines: Yahoo Finance ran both "oil prices jump to $105" and "Dow rises 622 points as oil prices fall after August CPI report" on the same page. The second is the one the settled contracts support.
Equities took the second reading too. The S&P 500 closed up 0.86%, the Dow up 0.98%, the Nasdaq up 0.96% — a rally on a day core inflation surprised upward and hike odds jumped, which is hard to explain by the inflation data and easy to explain by crude backing off. The longer-dated contract on a US invasion of Iran before 2027, at 15.5% on $65.6m of cumulative volume and by far the most liquid geopolitical market checked, was unmoved.
So what: near-term, the oil leg of the inflation story and the policy leg have separated. A Hormuz agreement removes the escalation premium; it does not change what the August core print already told the Fed.
4Hyperscaler borrowing now rivals the Treasury's at the long end
Three channels reached for the same fact and produced three numbers that do not reconcile, which is itself informative about how new the measurement is.
Nordvig, who says his team is "doing these calculations and trying to make them very precise", put it at parity on a flow basis: "we now have as much bond issuance by hyperscalers in the long end of the curve as the United States government" — and that against a US deficit running at 6% of GDP. Felix Prehn, on Felix & Friends, cited Goldman Sachs' Mark Wilson for a forward version: AI-related debt near $500bn by early August, roughly 15% of US investment-grade corporate borrowing, with $1.3tn expected next year, at which point a handful of technology firms would out-issue the US government. Ross Givens put this year's ratio at 70 cents of long bonds for every dollar the government sells, with $320bn issued and $1.65tn of off-balance-sheet commitments — chip purchase agreements and data-centre leases that are not yet counted as debt.
Parity, 70%, and 15%-of-corporate are three different measures of three different things, and none of the three sources shows their working. What they agree on is direction and that next year is larger than this one.
Martin supplied the mechanism in the plainest form. An investor facing a Treasury auction and a bond of similar duration on the same day, she said, may well take the corporate paper — "this is leeching demand out of government bond markets at precisely this time that people are worrying about the sheer levels of borrowing that governments have got to do." She reported smaller European governments rescheduling their own issues to avoid colliding with hyperscaler deals, and called it "one of the big underappreciated stories in global finance at the moment." US national debt has crossed $40tn.
Nordvig named the signal to watch: a large technology bond deal that struggles, gets downsized, or has to pay up. is the name he flagged as the weaker credit in the group, with volatile spreads; he noted had already paid extra to get its last deal done, and that order books which ran five-times covered in February were under two-times by July. Givens independently cited the same February-to-July deterioration.
So what: this is a source of upward pressure on long yields that no central bank sets and no Treasury buyback offsets. Martin's description of Bessent's programme — and a trader's line quoted by Prehn calling the buybacks "a firecracker in a hurricane" — is that it is small relative to what it is fighting.
What the sources recommended
Four calls came out of five videos, all four from one presenter, and all four are funds rather than companies — the other four videos named no buys and no sells, which for two macro interview shows and an options-positioning briefing is the ordinary outcome rather than a notable one.
Ross Givens laid out a four-fund allocation built on the premise that the cost of artificial intelligence is collapsing toward zero, so margin migrates from the sellers of intelligence to the owners of things it cannot make: , the equal-weight S&P 500, at half the portfolio, on the argument that cap-weighting concentrates a third of an index fund in seven names; , short-dated Treasury bills, at 20%; at 15%; and , a global natural-resources basket, at 15%. The caveat is his own and is part of the call: he framed it explicitly as what he would do "if I were 20 or 30 years further down the line with huge profits", not as general advice, and said of the incumbents that and the hyperscalers "could go higher" as readily as they could halve.
Two of his supporting observations are checkable and worth separating from the thesis. Software lost roughly $2tn of market value between September last year and March, trading at a discount to the rest of the index; and fell 18% in a single session on 19 June and is down more than 50% this year. Both are offered as evidence that the buyers of intelligence, not its sellers, have taken the damage so far.
Felix Prehn's material was directionally bullish on gold and silver but attached to no instrument, so it is not counted above. He relayed Goldman Sachs metals head Tony Kim describing the drop from January as "an elongated pause", with $4,000 as a floor and advice to clients to scale in on dips, and central-bank buying above 1,000 tonnes a year against roughly 3,500 tonnes of annual mine supply. Prehn said plainly that he does not share the market's expectation of a rate increase, which places him against both the venues in the first claim.
Three dated events test all of this inside a week
The FOMC decides on Wednesday 16 September, and VIX expiration falls the same day. Quarterly options expiration follows on Friday 18 September. Kochuba's measure puts roughly $2tn of delta-notional expiring, close to the September 2024 record of about $2.1tn and possibly past it by the time positions settle; he was careful to distinguish this from the $9tn figure Citadel has quoted, which counts every contract including far out-of-the-money ones at full share equivalence.
His positioning read, recorded before Friday's rally, is that index gamma has fallen close to negative while single-stock gamma stays positive, a combination that widens forward volatility. He named 7,600 on the S&P 500 as the level below which he would expect the VIX to reach 20 quickly, 7,350 as where positioning thins beneath that, and 7,700 to 7,800 as the upside into expiration if Iran de-escalates. Friday's de-escalation in the oil contracts happened after he spoke.
Further out, Polymarket prices Democratic control of the House after the 3 November midterms at 87.5% on $175k of daily volume and $866k of resting liquidity, while the Senate is a coin flip at 52.5% Democratic. Martin's interest there is not the result but what follows it: whether a president perceived as having a year left changes the behaviour of other governments. A UK budget lands in the same window, with gilt borrowing costs already at their highest since 1998.
