1A core inflation print of zero bought no rate cuts — the hike simply moved to December
Core CPI printed flat month on month, a good deal colder than expected, and a cool PPI followed the next day. Joseph Wang, on Fed Guy, says that pairing all but removed the odds of a July hike and left core PCE looking acceptable. What it did not do is buy any easing. One hike is still priced by December — postponed, not cancelled.
The prediction-market board puts it more starkly. Polymarket's contract on how many Fed cuts land in 2026 is the deepest thing on the board at $43m of volume and roughly $3.3m of liquidity, and it prices no cuts at all this year at 84%, against 12% for a single cut. A separate contract on whether the Fed hikes at some point in 2026 sits at 54% on $4.2m. The venue with the most money behind it is not arguing about how much easing is coming; on balance it is leaning the other way.
The official commentary explains the asymmetry, and the data gives the Fed room to sit still. Chair Warsh reiterated the price-stability mandate on the Hill, and Governor Waller — whom Wang singles out as the influential voice — said on Monday that a hot core print would have him backing a July hike. It came in cold, so that question answered itself, but neither official offered a reason to price relief. Meanwhile bank earnings landed solid: Bank of America reported customer spending up 5% year on year with credit quality intact and no rise in charge-offs, and JPMorgan said the US consumer is fine across the board, with no K-shape visible in its own data. A credit or labour crack is usually what forces a cut, and neither is in this week's numbers.
Gold is where the argument shows up in a price. It had its biggest weekly drop since June, then bounced 0.67% on Friday as a safe-haven bid returned, with spot around $4,018. Polymarket's July gold market, on $591k of volume, prices a 35% chance of tagging $3,900 before month-end and 13% of $3,800 — the skew is to the downside. Higher real yields press on gold in a way a geopolitical bid can offset for a session but not obviously for a month.
Two things stop this being settled. Wang's own framing is that the path stays data-dependent and the calendar is awkward — November midterms make a September move unlikely — and that an equity decline would take the hikes out of the curve entirely. And the administration is telling a different story about the same data, repeatedly touting June CPI as the largest single-month fall in prices in six years. That is a disinflation message the rates market is simply not trading. On this evidence the route to a cut runs through an equity drawdown, not through the inflation prints.
2Semiconductors sold off on good news, which points at positioning rather than demand
TSMC fell 7% on Friday and sold off as well — both on upbeat AI demand, not in spite of it. Wang flags the pattern itself as the warning: when good news is met with selling, the information is in who already owns the stock rather than in the news. was dragged along as the complex's leader, and sat at the epicentre, memory being where the volatility concentrated.
The second blow was what Wang calls a new DeepSeek moment — Kimi, an openweight Chinese model from Moonshot, arriving at roughly frontier-comparable quality. The leverage attached to that trade was visible in Korea, where the Kospi went parabolic and then tumbled, and more than a million Korean accounts were margin-called on Friday.
Andy Constan of Damped Spring, on Excess Returns, had flagged the setup in a report dated around 28 June, and his evidence was the options tape rather than the price chart: parabolic moves, very high single-stock volatility, and an inverted skew with calls bid richer than puts — "bull market for calls, bear market for puts," as he puts it, with hedging demand absent. That inversion is normally a bottom signal, panic reaching for calls after a fall. Here it appeared near the highs after a long rally, which he reads as a speculative frenzy peaking instead.
What he did about it is the part worth keeping. He sold call volatility and did not short the market — a non-directional expression, and he is explicit that it is not a top call. Since the report, he says call demand has vanished, put demand has returned and skew is re-normalising, which fit the trade; but realised volatility is still extreme, with Hynix and Samsung printing limit-up and limit-down days and 10% daily moves.
His structural case is an arithmetic one about earnings rather than a chart. Nearly all AI capex funnels into a handful of names — roughly three in memory, three or four in compute chips, one raw-silicon maker, plus the clean-room build enablers — and there is not enough GDP for every one of them to earn the returns already priced in. Even on aggressive productivity and capital-share assumptions he cannot make the built-in earnings work, so on his numbers some of these names get no pie at all. is his parabolic example, up three to four times.
Counter-evidence sits on that same ticker: Intel's 18A process was reported at an 85% yield, which is a manufacturing result rather than a sentiment one. And Constan's own nuance cuts against his conclusion — if usage migrates to cheaper open-source models, that can help semiconductor demand rather than hurt it, because inference compute rises. So the Kimi event is genuinely ambiguous for chips, and he says so.
Corroboration for the frenzy reading arrives from an unrelated seat: Goldman Sachs' equities revenue was up 50% year on year on IPO and leverage activity. Political framing, meanwhile, points the other way from the tape — the administration praised TSMC's US leading-edge fab buildout as a reshoring win in the same window the stock fell 7%, and called data centres "money machines" while attacking New York's governor for terminating them. Policy support and price action are not the same signal.
Two channels, not eight: the weekend feed carried fresh episodes from Fed Guy and Excess Returns only, so the agreement between them is two voices rather than a chorus. What both describe is a complex where good news has stopped lifting prices — and positioning unwinds faster than fundamentals change.
3The escalation around Iran resumed against a strategic reserve near multi-year lows
Wang's read is that the war is back on: a military buildup in the region, refuelling tankers positioned in Israel, and Iranian strikes on US bases. He cites Professor Pape reading the US targeting pattern as preparation for a ground invasion.
The supply side is what makes this a market event rather than a headline. The Strategic Petroleum Reserve is near multi-year lows, refinery problems raise the risk of diesel, jet fuel and gasoline shortages, and spot oil is already rising. A thin reserve means less shock absorber.
Polymarket prices the risk as live rather than tail. Its contract on Iranian military action against a Gulf state before the end of July sits near 56% Yes on $1.8m of volume, and the deeper board around the conflict carries real money — $44m on whether the US invades Iran before 2027, $6.9m on the timing of the next round of US–Iran talks, $4.4m on a full Iranian airspace closure. News flow runs the same way: Iran has already struck targets in Gulf states and Jordan.
The disagreement here is chronological. The administration's supply-security message — the Strait of Hormuz "open to ALL Shipping" — was posted on 14 July, before the weekend re-escalation. The reassurance and the risk describe different weeks, and only one of them has been updated. Which is the consequence worth holding: the oil exposure here is a refining-and-reserve risk rather than a crude-headline risk, and it is the one the reassurance does not address.
The options cushion comes off midweek, into the mega-cap earnings block
Wang's outlook for the week ahead is more geopolitics and less support from the options complex once Wednesday's VIX expiry passes, which is his stated reason for turning a bit more cautious. That lands directly on the second claim above: if the semiconductor weakness is positioning rather than demand, then withdrawing the options-market support is exactly the condition under which positioning moves prices.
The mega-cap earnings block arrives in the same window, and this season has already produced violent single-name dispersion — plunged 22% on its 14 July print, rose 6% on a beat two days later, and and were both trading lower in Friday's pre-market. Earnings are the one input that can separate the two readings: a beat that gets sold is a positioning story, a miss is not.
