1The consumer contracted and the market still thinks the Fed's next move is up
The Census Bureau's advance retail sales report landed this morning and it went the wrong way. July sales came in at $763.6bn against $768.1bn in June — a fall of about 0.6% on the month. Strip out food services and the drop is steeper: $660.0bn against $665.1bn, roughly 0.8%. This is not a soft patch inside a rising trend. It is a decline in the largest single component of American demand.
It arrives on top of a labour market that already stopped adding jobs. July payrolls fell by about 23,000, to 158.858 million from 158.881 million in June. The unemployment rate went the other way, down to 4.1% from 4.2% in June and 4.3% in May — which tells you the drop came through people leaving the labour force rather than through hiring. And inflation is not the constraint it was: the July CPI index rose 0.074% on the month, an annualised pace under one percent.
Weak demand, no job growth, cool prices. The textbook response is easier policy. The market is pricing the opposite. Polymarket's September Fed contract — $33.0m of volume against $3.5m of resting liquidity, the deepest thing we checked today — puts no change at 74.5%, a 25 basis point increase at 24.5%, and every cut outcome combined at 1.6%. A rise is priced at roughly fifteen times the odds of a cut, three days after a jobs report that lost jobs.
The retail print did register. Over the last 24 hours the hike leg fell 4.0 points and no-change rose 4.0. So the contraction moved money — it just moved it from "hike" to "hold", and never as far as "cut". Whatever is holding the hawkish side up survived a shrinking consumer intact. The next section is where it lives.
2The market is pricing a frozen Iran conflict — no war, no deal, and a chokepoint that stays shut
The headline version of today is escalation. Reuters reported oil climbing on tanker attacks and on competing US and Iranian claims about who controls the Strait of Hormuz, with transit traffic capped and slowing further after Washington threatened more economic pressure. Treasury yields rose on the sanctions threat. Scott Bessent told Newsmax the US would hit Iran with "unprecedented economic isolation" — a clip the President reposted to his own feed this morning.
The prediction markets tell a more specific story, and it is not escalation. The Israel–Iran ceasefire contract for continuation through 15 August sits at 99.3%, up a point on the day. The shooting has not restarted. But a US–Iran final nuclear deal by 31 August prices at 1.05%, down 0.4 on the day, on $14.8m of volume and $1.35m of liquidity; the 30 September leg is only 4.5%. And Hormuz traffic returning to normal by 15 September is at 6.5%, down 2 points today, on $140k of volume against $129k of resting liquidity — thin enough to read as direction rather than precision.
Three contracts, one configuration: no war, no deal, and a waterway that stays disrupted. That is the arrangement that keeps a persistent oil risk premium in the price without an escalation spike to clear it. Sanctions tighten supply; they do not restore transit. A premium that sits there is an inflation input a central bank has to carry in its forecast, and it is the most plausible thing propping up the 24.5% hike leg from the section above while the consumer shrinks underneath it. Note what the market did today, though: it took the hike down, not up. The disinflationary read from a contracting consumer is winning at the margin — the oil premium is holding the floor under the hawkish case, not driving it higher.
3The last unfired crash signal is a credit spread sitting near a record low
Ross Givens spent his video walking a fourteen-item checklist of markers he says preceded both the March 2000 and October 2007 tops, and counts thirteen as already triggered. The specifics are worth keeping whatever you make of the frame. He cites Morgan Stanley putting AI-related debt issuance near $570bn this year, with hyperscaler bonds at $225bn by mid-year. He cites Moody's counting $1.2trn of off-balance-sheet AI commitments, with disclosing $329bn of leases not yet commenced against $93bn a year earlier while extending assumed data-centre life from 15 years to 25 — a change that lowers reported depreciation without anything physical changing. He puts five-year credit default swaps at above 215 basis points in July against 145 in January, and margin debt at $1.53trn, up 51.5% in a year.
The unfired one is the high-yield spread — what the riskiest borrowers pay above Treasuries. It is 2.7 percentage points now, near the tightest it has ever been. His trigger is 3.5 points held for a month, not merely touched. In 2007 that spread sat at 2.3 in spring and reached 4 by August, four months before the S&P peaked. His own caveat travels with the call and matters: he says explicitly he is not forecasting a crash tomorrow, and points out most of these markers were flashing in 1999 while the Nasdaq doubled before topping. He also pitches a paid subscription service three times inside the video, so there is a funnel behind the framing.
Two of his inputs get independent corroboration from a source with no bearish position to sell. On The Compound and Friends, Todd Sohn of Strategas put levered ETF assets at a $200bn record with $500bn in notional, and said the demand for leverage came straight back after July's shakeout: "we're not through this yet". Josh Brown, on the same episode, described this week's Nasdaq round table — Jensen Huang alongside David Solomon, Apollo and Blackstone announcing a $500bn partnership — as "a round table of circular financing". The leverage and the vendor-financing structure show up in flow data and on CNBC, not only in a bear's deck. What has not shown up is the credit market charging more for risk, and on this record that is the signal that has historically dated the top.
4Quantum's best news quarter moved no share prices, and the packaging is outrunning the assets
Felix, on Felix & Friends, made the sharpest version of a mispricing argument today, and the gap he describes is real and checkable. The last three months brought Google's chief executive saying on camera that quantum is where AI was five years ago; an 11 August announcement that Oracle and Quantinuum are physically installing a 98-qubit Helios machine inside an Oracle cloud data centre; and roughly $2bn of US government money in May taken as equity stakes across nine companies, about $1bn of it into , with and around $100m each. Against that, he puts down 2% over the stretch, up 2% and up 6%, with all of them still 45–70% below last October's highs.
The one argument with a market mechanism under it is power. He cites the Helios machine drawing 60 kilowatts against 16,000 to 39,000 for a leading supercomputer. Electricity is the binding constraint on the AI build-out, so a machine that competes for cloud budget at a rounding fraction of the draw is a different proposition from a research curiosity. He also notes reported revenue up 287% year-on-year to about $80m in the quarter against a backlog near $500m, still lossmaking, and that bought a chip foundry to bring manufacturing in-house.
His hedges are part of the call and several bite. He is cautious on and wants it above roughly $20 before buying — it traded at $18.90 this afternoon, below his own threshold, so his stated condition is unmet as of today. He demoted from his list while saying outright that owning it is not daft. And he repeatedly promotes a webinar and a stock-screening app through the video, so the delivery has a funnel attached the same way the previous section's did.
The packaging is moving faster than any of this. Sohn put 2026 ETF launches near 900 year-to-date against roughly 1,100 for all of last year, and described a Y-Combinator-backed newcomer launching thirty thematic funds at about 35 basis points against incumbents at higher fees. The detail that lands hardest: five photonics ETFs now exist against roughly three photonics stocks, and a Roundhill launch took more volume in four days than the other four combined. Brown's read was that the vehicles will keep coming and the audience will tire: "people got bored of that. They're going to get bored of this." A sector where the news improved and the price did not is either mispriced or the news does not yet reach the cash flows. Nothing gathered today settles which.
What the sources recommended
Six single-name buys came out of four videos, and all six came from one person in one video — so this is one investor's sector allocation, not a consensus. There were no sell calls at all.
Every buy is a quantum computing position, and Felix tiered them explicitly by risk rather than presenting them as equivalent. His lower-risk tier is the incumbents that carry the theme inside a working business: , which took about $1bn of government money and is putting a further $1bn into a 300mm quantum wafer plant in Albany; , whose Willow chip started this cycle and whose error rate falls as the chip scales; and , whose majority stake in Quantinuum is the machine Oracle is now installing — his argument being that you own a profitable industrial and get the quantum business inside it. His middle tier is , the pure play with actual revenue, which he paired with an instruction to size it carefully because it is still lossmaking. His speculative tier is and , on which he was blunt: "you're going to lose most of it", so position small. Two names he declined to add: , where he wants to see above roughly $20 first and the stock sits at $18.90, and , which he removed from his list while saying holding it is defensible.
Elsewhere the discussion stopped short of calls. Brown argued should trade 30–50% higher given the Lakers changing hands near $12.4bn against a roughly $10bn market capitalisation, with the Rangers spin-off from the Knicks due to complete by the end of October — then said he had missed it and only might buy, which is why it is not in the buy list. Sohn flagged buying Innovator and NEOS as a bet on buffered and option-income products for an ageing investor base, and named , , , and as the energy names on a best-in-market screen, with the observation that energy's beta to the index has collapsed as the sector shrank to about 2% of it — making it a hedge rather than a swing.
The consumer claim gets three read-outs next week
Section one rests on one month of retail data. Three of the largest US retailers report before this time next week and will mark it to market: before the open on Tuesday, then and before the open on Wednesday, with the same day. If the July contraction is a genuine turn in demand rather than a seasonal artefact, it should show up in their guidance rather than only in their quarters. , and report across the same stretch, which reads on Chinese consumption rather than American.
Two live policy items sit underneath. A 100% tariff on imported drones was reported today, which is what moved the drone names in the pre-market. And the Court of International Trade struck down the de minimis import exemption, a ruling the President posted about directly — that one reaches parcel volumes and cross-border e-commerce economics rather than any single listed name.
