The Dot Plot Nobody Priced

TO: Desk
FROM: Rates & Macro
RE: Wednesday's hike, and the thing everyone's pretending not to have read
CLASSIFICATION: Internal, but you'll leak it anyway

Let's get the theater out of the way first. The Fed hiked 25 basis points Wednesday, 12-0, to 3.75%-4%, the first increase since July 2023. Markets had it priced at 90-93% for a week. Nobody on this desk should have made a dollar off the headline number. If you did, that's a coin flip you're mistaking for a thesis.

Here's what you should have made money off, and mostly didn't: the dot plot.

Sixteen of eighteen FOMC participants penciled in at least one more hike this year. Four of them want two more. Warsh himself declined to submit a dot — a chairman abstaining from his own committee's forecast, which is either admirable restraint or the tell of a man who doesn't want his signature on a number he might have to eat in December. Take your pick. Median 2027 projection jumped from 3.4% to 3.9%. Read that again. Not "rates come down next year." Rates stay higher than the Fed itself was penciling in three months ago, with the first cut not even shown until 2028. That's not a pause-then-cut cycle. That's a committee telling you, in writing, that this hike is the first of several, and that the destination is a policy rate that would have been unthinkable in a 2024 base case.

The Dow didn't care Wednesday afternoon — down 631 points, Goldman and IBM leading the bleed, S&P off 0.4%, Nasdaq flat. Fine. Knee-jerk. Except then Thursday happened, and the S&P ripped back to 7,596, up 0.59%, erasing most of the damage in a single session on a day when there was no new information except twenty-four more hours to digest a hawkish forecast. If your model says "hawkish surprise, sell the rate-sensitive stuff, hold the cash-generative stuff," somebody forgot to tell the index. We are one day removed from the Fed telling us rates stay elevated through 2027, and the tape's response was to buy the dip like it was a Waller soundbite.

Three things are keeping this from being a full-blown repricing, and all three are borrowed time.

One: the 10-year sitting at 5.01-5.04%, a 2007-level yield, has already done a chunk of the Fed's work for it. The bond market moved before the Fed voted — the hike Wednesday was closer to ratification than surprise, and equities have had weeks to adjust to a 5-handle on the long bond. The dot plot is new information about the path; the level itself is stale news to anyone who's been watching the 30-year sit above 5.3% since last week.

Two: oil gave the market cover. Crude pulled back from its highs Tuesday on a surprise build in US inventories, softening the inflation-pass-through narrative for exactly the 48 hours the market needed to stop pricing in a second Hormuz-driven CPI shock. That's not the same as the supply problem being solved — Saudi's East-West pipeline is still down, Libyan fields are still offline, Russian refining capacity is still getting drone-holed — it's a data point traders used as permission to stop panicking, not evidence the panic was wrong.

Three: gold. It slid to $4,307 on Tuesday, down nearly 2% as yields and oil both firmed, then clawed back above $4,300 Wednesday as the actual hike underwhelmed relative to the pre-meeting fear. That's a market treating "as expected" as good news even when "as expected" now includes a forecast for materially more tightening than existed a quarter ago. Gold's relationship to real yields is supposed to be inverse. Watch what it does over the next two weeks once the dot plot actually gets modeled into forward curves instead of skimmed off a press release.

The internal read: nobody has actually repriced 2027 yet. The market spent Wednesday and Thursday trading the binary — hike or no hike — and largely ignored the SEP, because a scatter chart of eighteen anonymous dots is harder to trade than a single percentage figure at 2pm. That's a mistake. The Fed just told you, with about as much clarity as this institution ever manages, that it thinks the terminal rate is meaningfully higher than consensus assumed in June, that Warsh doesn't want his name on the number but isn't fighting the committee's direction, and that nothing resembling relief shows up on their own timeline until 2028.

Position accordingly, or don't, but don't tell me Friday you got caught by something that was published Wednesday in eighteen dots nobody bothered to count.

END MEMO