1A September rate cut has been priced out — the argument is now hike or hold
The September Federal Reserve decision is no longer a question of how much easing. On Polymarket, a 25bp cut prices at 0.45% and a 50bp cut at 0.15%, against a 25bp increase at 53.5% and no change at 46.5%. That is real money, not a thin quote: the hold and hike legs carry roughly $821k and $812k of liquidity, and the hold leg alone turned over $629k in 24 hours. The hike leg rose 2.0 points on the day; the hold leg has gained 8.0 points over the week, so the two are converging rather than one running away.
The bond market agrees on direction. The 10-year Treasury yield printed 4.78% on 4 September, up from 4.77% the day before and holding a 4.77–4.79% band all week, against an effective fed funds rate of 3.63%. The 2s10s spread sits at +0.41, so the curve is steepening rather than inverting — the shape you get when the market prices inflation risk, not recession.
The hard labour data supports the hawkish read on its face. August payrolls came in at 159.08 million, up 162,000 on July, with unemployment steady at 4.1% for a second month after 4.2% in June.
Katherine Anne Edwards, a labour economist speaking on Prof G Markets, reads the same report the other way. Her argument is that the headline is carrying a report that is weak underneath it: the share of people unemployed six months or longer is "way too high to call the labour market good", wage growth is "now slipping behind inflation", and there are over a million fewer prime-age men in the labour force than a year earlier. She puts the rise in the number of unemployed over three years "on par with a kind of like mild recession" — while saying plainly that this does not mean a recession is under way. Her framing of the monthly number is that it is "closer to betting than it is a report on the labour market's quality."
Jared Dillian, on Excess Returns, takes the other side of the inflation question outright: "at the moment, I'm not worried about inflation. I think inflation is coming down."
So the split is clean and unresolved. The market prices a hike as marginally more likely than a hold; an economist reading the same release sees a labour market weak enough to argue the opposite; and core PCE rose 0.25% in July, an annualised pace near 3%, which is warm enough to keep the hawks honest. What would settle it is the next inflation print, not another jobs headline — both sides here already agree the monthly payroll number is too noisy to carry the argument.
2Crude is pricing an escalation that the ceasefire market keeps de-rating
Oil is the day's clearest move, and the prediction market explains it better than the headlines do. Polymarket's contract on a US–Iran effective ceasefire beginning by 11 September fell to 60.0%, down 10.5 points in 24 hours on $43k of liquidity. The 30 September leg fell to 72.0%, down 8.0 points, on $48k of liquidity and $87k of 24-hour volume. Both legs are still up on the week — 16.0 and 13.5 points respectively — so this is a sharp one-day reversal inside a slower improving trend, not a collapse.
The tape moved with it. closed up 2.82% on 8 September and is up 9.2% over six sessions, from 133.72 to 146.01. added 1.09%. Reuters had crude "heading for $100" overnight, and Ed Elson put Brent near $98 on Prof G Markets. The escalation behind it is specific: US forces destroyed five Iranian oil carriers after attempted missile attacks, Iran says it captured a US submarine drone in the Strait of Hormuz, Houthi attacks disrupted Saudi energy facilities and wounded 73 people, and the US Treasury added sweeping new aviation sanctions.
One aggregator headline in the same window reported oil falling below $80. It is contradicted by the price itself and by every Reuters item on the wire, so treat it as stale.
Against all of that sits the President, who wrote on Truth Social that oil prices "will drop precipitously… when we WIN the war with Iran", naming three dollars a gallon and "ultimately, below Two Dollars a gallon". That is a forecast conditional on an outcome the ceasefire market just marked down. The market and the claim can both be honest — they disagree about the probability of the condition, not the consequence. A ceasefire stops the escalation premium building; only a durable Hormuz settlement returns the disrupted barrels, and the contract on an Iran–Oman Hormuz management agreement by 30 September sits at 26%, down 6.0 points on the day.
3AGI was declared, and none of the effects that were supposed to define it showed up
OpenAI released GPT6, called Astra, on Friday. It was trained on more than 100,000 GPUs at the Stargate data centre in Texas, the largest training run in the company's history, at two and a half times the cost of its predecessor. Nvidia chief executive Jensen Huang said artificial general intelligence had "arrived"; OpenAI president Greg Brockman wrote "Welcome to the AGI era."
Gary Marcus, emeritus professor of psychology and neuroscience at NYU, speaking on Prof G Markets, argues the label has been detached from anything testable. His mechanism is what he calls benchmark-maxing: models are trained heavily on benchmarks, score well on them, and then underperform in real use. His test from 2014 — that a system should watch a film and describe what happened as well as a high-school student — is still unmet twelve years later, which he offers as evidence that he has not moved his own goalposts.
The sharpest line is about incentives rather than capability. Marcus notes Huang had already declared AGI on a podcast six months earlier, and that the cost of such announcements "is not just zero — it's that it drives up the stock price." Sam Altman, asked directly on another podcast whether AGI is here, called it "at best… a very poorly defined term" and answered "sort of, close at least" — a hedge Marcus reads as avoiding any commitment.
Marcus is a long-standing critic of this industry, and this was an interview that put no counter-case to him, so it is one side argued well rather than a debate. The testable part of his claim is the part worth keeping: if the strong definition held, wages would fall and unemployment would rise. Neither is happening — August unemployment was steady at 4.1% — which means either the definition has moved or the milestone has not been reached.
4The S&P 500 has become a tech index, and that is the risk inside the passive trade
Jared Dillian, on Excess Returns, is not making a market call — he says so directly: "I'm not going to make any stock market forecasts on this podcast." His argument is structural. Index funds were 2% of assets under management when he opened his first Vanguard account in 1997; they are close to 60% now. Roughly 45% of the S&P 500 sits in its top ten holdings, and in his description those are "pretty much all tech stocks", so "the S&P 500 has turned into really… a tech index." The diversification that was the product's original selling point is the thing concentration has eroded.
His stated risk is a crowding one: with that many holders in the same position, a simultaneous exit produces a liquidity stampede, which he says is what the 35% drawdown inside a month during the pandemic actually was. He is careful about the timing claim — "anytime concentration gets to these levels it's usually at or near a top" is as far as he goes, and he pairs it with an admission against interest, that his book "is coming out at a very bad time" because an eighteen-year run has left few people receptive.
His alternative is an allocation, not a trade: 20% each in stocks, bonds, gold, cash and real estate. On his own figures to 1 January 2026 it carries a Sharpe ratio of 0.6 against 0.7 for the S&P 500, with standard deviation of 8.22% against 17.04%, and a worst year of −11.8% in 2022 against the index's −36.55% in 2008. Note the Sharpe ratio is lower, which is the honest version of the pitch: the case is the volatility halving, not a better risk-adjusted return. He also flags the one regime that breaks it — rapidly rising interest rates, which hit bonds, stocks and gold together — which is precisely the regime the September Fed market is currently pricing at better-than-even odds.
What the sources recommended
Three videos produced one fresh single-name buy between them, and the two macro conversations produced none at all — a thin day for calls, and worth reading as such.
The one call is Ross Givens buying , Hinge Health, which he took as a couple of hundred shares at 91.84 with a stop at 84 — a stated risk of about 9%. His case mixes a seasonal argument with a company one. Seasonally, he cites healthcare as up 100% of the time from July to December of midterm election years, inside a fourth quarter that is higher 86% of the time. On the company, he cites second-quarter sales up 53% to $213 million, $100 million of free cash flow in the quarter, full-year guidance raised to roughly $860 million, a $300 million buyback and an acquisition in gut health, at 69 times trailing and 36 times forward earnings. He attaches his own caveat: "don't bet the farm on this", naming the Iran situation specifically.
He said the ticker aloud as "HNG", which does not exist. The company is Hinge Health, and the symbol is on the NYSE — confirmed by the $7.42 billion market capitalisation he quoted, which matches.
He also updated an existing position rather than recommending it: still holding, up about 14.8% over three sessions, and planning to sell some into 300–320. That is a position update with a trim attached, not a fresh call.
Dillian's asset-allocation case for gold, bonds, cash and real estate is a portfolio construction, not a set of single-name calls, and is recorded as such. No source recommended selling anything.
The escalation trade gets marked to market at Wednesday's open
The ceasefire contract expires on its 11 September leg in two days, which makes the next 48 hours a direct test of the 60% the market is now paying. If the strikes on tankers and the Hormuz drone seizure continue, that leg falls further and the crude bid holds; a de-escalation headline reverses both at once.
reports after the close, a defence and drone manufacturer reporting into the middle of exactly this conflict, which makes its guidance more interesting than its quarter. and report before the open as consumer reads, useful against Edwards' claim that wage growth is slipping behind inflation. An Nvidia event is scheduled for 10 September, one day after the AGI argument above.
The inflation print, not the next jobs headline, is what settles claim one — and nothing on this week's calendar delivers it.
