The 85% Problem

Somewhere between Thursday's CPI print and Friday's close, the market decided two mutually exclusive things at once. It decided the Fed is now 85.5% likely to hike on September 16 — up from 56% a week earlier, up from a coin flip the week before that. And it decided this was cause for a 509-point Dow rally.

Sit with that for a second, because nobody else seems to want to.

August core CPI came in at 0.3% month-over-month, hotter than the 0.2% consensus, with headline at 0.4%. That's not a rounding error. That's the print Collin Martin at Schwab correctly flagged as giving the Fed nothing resembling confirmation that disinflation is back on track. And the response, after a brief yield wobble, was stocks grinding higher into the weekend as oil eased off its Hormuz-driven highs and long-end yields took a breather from their multi-year highs. The S&P closed up 0.9% Friday. The Nasdaq 100 did the same. Four straight down sessions, reversed, on a week that should have been unambiguously bad news for anyone holding duration or growth multiples.

Here's the thing about an 85.5% probability: it isn't a forecast anymore, it's a foregone conclusion wearing a forecast's clothes. When CME FedWatch moves that far that fast — nearly 30 points in under two weeks — the number stops describing uncertainty and starts describing consensus. And consensus, at 85%, means the hike is already in the price. Which is exactly the kind of environment where the actual announcement becomes almost irrelevant and the post-meeting press conference becomes the whole show.

That's the bet embedded in Friday's rally, whether the people making it would phrase it that way or not. Nobody is pricing "maybe no hike." They're pricing "hike, plus whatever Kevin Warsh says next that sounds less hawkish than what he said at Jackson Hole." That is an extraordinarily narrow needle. Warsh spent August building a reputation for the opposite of dovish surprises — "not broadly restrictive," "we have work to do," a Fed chair who has made a project out of not indulging markets that expect him to hand them their next trade. A market betting on a dovish pivot in his press conference tone, five days after his own committee data forced a hike, is a market betting against the guy's entire stated identity.

And yet here we are, with Bitcoin holding in the mid-$70,000s to low-$80,000s — essentially flat on the week, occasionally poking above $81,000 — instead of cratering the way a "priced-for-certain rate hike into hot inflation" backdrop should produce for a non-yielding asset. Gold, which spent August putting in its best month since 1999, has spent September giving a chunk of that back, currently sitting well off its ~$4,600 peak. Both assets are technically consistent with a hiking cycle. Neither is behaving like markets that actually believe the tightening thesis all the way through. They're behaving like assets waiting for someone to blink.

The honest read is that three different trading desks are running three different models of the same week, and none of them are being reconciled because there's no need to reconcile them until Wednesday. The equity desk is pricing "hike now, cut path resumes later" — a garden-variety hawkish-pause narrative dressed up as a hike, on the theory that one 25bp move that's already fully discounted removes uncertainty rather than adding restriction. The rates desk is pricing what it's actually seeing: a 10-year that touched 4.818% two weeks ago, its highest level since November 2023, on the entirely reasonable view that a Fed hiking into 3%+ core PCE and Hormuz-driven energy prices is a Fed that isn't done. And the crypto and gold desks are pricing neither of those stories cleanly, because both assets have spent 2026 failing to behave like the macro hedges or risk proxies they're supposed to be, and have instead become referendums on their own idiosyncratic flow — ETF creation baskets, central bank buying programs, leverage unwinds — that happen to correlate with rates sometimes and not others.

None of these three desks is wrong on its own terms. What's wrong is the implicit assumption, priced into Friday's rally, that they're all going to turn out to agree with each other by Wednesday afternoon.

The Fed's own dot plot lands the same day as the decision — this is a Summary of Economic Projections meeting, so Warsh doesn't get to hike and then hide behind vague forward guidance. He has to hike, if he hikes, and simultaneously show the committee's hand on where the terminal rate sits and how many more of these are coming. There is no version of that combination — a hike, a hawkish chair, and a published dot plot in an environment of $90+ oil and 3%-plus core inflation — that reads as good news for a 10-year note or, transitively, for the multiple the market is currently willing to pay for the Nasdaq. The only way Friday's rally makes sense in hindsight is if the SEP shows a dovish enough 2027 path to offset the September move itself — the classic hike-and-cut-the-guidance combination that lets everyone declare victory simultaneously.

That is a genuinely plausible outcome. It's also not the base case implied by a Fed chair who spent Jackson Hole explicitly warning markets against expecting exactly that kind of accommodative signal. The market has three days to decide whether it actually believes Warsh's rhetoric, or whether it's simply assumed, without much evidence, that inflation-fighting conviction always bends the moment a hike is banked. History offers examples of both. It does not offer a way to know in advance which one this is — which is precisely why pricing 85.5% certainty on the vote and near-total complacency on everything that comes after it looks less like conviction and more like a market that has stopped doing the second half of its job.