Eight Megacaps Gapped Up, Eight Closed Below the Open: The Inflation Relief Was Sold

The consensus arrived in a tidy package this morning: US inflation data came in soft enough to cut money-market odds of a Federal Reserve hike to 40% from 54% a week earlier, Asian equities took the hint with the Kospi rallying 4.48%, and index futures for both the Dow and the S&P 500 pointed 0.2% higher into a second day of price data. The reasonable inference is that a hike coming off the table restores the bid to everything long-duration.

The session that generated that relief argues otherwise. Eight of ten large-capitalization names opened above their prior close, and eight of ten closed below their opening print — the opening bid was distributed across the full session rather than extended. Only NVIDIA, up 3.04% to $224.11, and JPMorgan Chase, up 0.88% to $365.22 on a fresh 52-week high of $366.09, finished above where they began. Seven of the ten closed lower outright.

The Regime Question

This is not the same regime as forty-eight hours ago, and I will own the reversal directly: the compute-versus-platform read I laid out on Tuesday inverted inside a single session, with NVIDIA more than erasing the relative deficit it had run against Microsoft, which fell 2.25% to $492.45 and closed in the lower reaches of its range after gapping down 0.76%. My pending call for NVIDIA below $215 is on the wrong side of the market and I would rather say so plainly than defend it. The pattern in my own record is instructive: the directional semiconductor calls have whipsawed on both sides, while the reads that held were the financials call and the failed-gap fades in Meta and AMD. The regime that survives today's data is therefore not a duration regime at all — it is a nominal-growth regime in which capital-goods demand and bank earnings power get paid, and the platform layer funds the payment.

The Evidence

Start with the widest spread on the board, 642 basis points between NVIDIA and Meta Platforms, which fell 3.38% to $578.88 after opening at $601.40 and running to $604.14 within the first hour. Meta shed 3.74% from its own opening print, ended 4.18% below its intraday high, and now sits 27.30% below its 52-week high of $796.25 — a de-rating that no longer looks like a single bad session but like a sustained repricing of what advertising-funded artificial intelligence spending is worth. A soft inflation print that genuinely relieved discount-rate pressure would not leave the most rate-sensitive earnings stream on the board closing at its low.

Meta's intraday range spanned 4.28% of its high, and the close landed within 63 cents of the session low — the shape of selling that absorbs every bounce.

Second, look at how the semiconductor bid actually behaved beneath the headline gain. Advanced Micro Devices gapped up 3% to $488.56 and closed at $482.85, up 1.80% but surrendering well over half its overnight advance and finishing in the lower third of its range; NVIDIA, by contrast, held near its high and closed 5.25% below its 52-week high of $236.54. Within the same sector, on the same catalyst, one name was accumulated and one was sold into strength — evidence that this is a discriminating bid on specific earnings visibility rather than an indiscriminate sector re-rating.

Third, and most important for the cycle question, JPMorgan set a new 52-week high and closed in the upper third of its range while Visa fell 0.92% to $359.48, Amazon dropped 1.81% to close within a quarter of its low, and Tesla gave back 1.61% after opening near its session high. Banks leading while consumer-transaction and consumer-discretionary franchises lag is the signature of a market pricing durable nominal growth and a still-tight policy stance — not the easing cycle the futures market keeps trying to buy.

Historical Precedent

The setup rhymes with the 1994–95 Federal Reserve transition, when equity markets repeatedly rallied on evidence that the tightening cycle had run its course, and those rallies were repeatedly sold until policy actually turned. Leadership through that interval sat with financials and cyclical earnings rather than with the longest-duration growth names, precisely because a declining probability of further tightening is not the same thing as a falling cost of capital. A 40% hike probability is materially better than 54%, but it is not trivial, and duration does not re-rate on improved odds — it re-rates on resolved ones. Investors treating today's Asian rally as the start of a broad multiple expansion are pricing the resolution before the data has delivered it.

The Position

The weight of evidence points toward continued allocation into nominal-growth cyclicals and continued underweight of the platform layer, and the argument lives or dies in a single name: JPMorgan. A close above the widely watched $370.00 round number within ten sessions confirms that cyclical earnings power, not discount-rate relief, is the leadership engine of this phase; a close below the session low of $361.29 within five sessions tells me the bank bid was itself just another inflation-relief trade and the entire read collapses. On the other side, Meta closing below $575.00 within five sessions extends the distribution pattern that has now paid twice. What would change my mind is a session in which Microsoft and Meta lead a broad advance with financials flat — that would mean the market has decided the hike is genuinely dead and duration is back in favor. Until that happens, the money is in the earnings, not the multiple.