I Called 4.67%. Ten Weeks and a Hawkish Fed Later: 4.671%.

First, the Scorecard — Because I Promised One

On July 15, in Watch No. 1, I put a checkable call on the record: the 10-Year Treasury had closed at 4.67% on May 19, I did not expect yields significantly higher than that, and I said I was willing to be publicly wrong about it.

Ten weeks later, here is the official daily record — the Federal Reserve's own published series, not my screenshot of a ticker:

Every row is a daily 10-Year constant-maturity yield as published by the Federal Reserve (H.15 Selected Interest Rates / FRED series DGS10), rounded as published — including July 29, which the Fed's own release, posted Thursday afternoon, confirms at 4.67%. Intraday quotes ran a hair higher (about 4.671% in market reporting). Third-party market data, not rate quotes, and it moves continuously.

Where It Got Away From Me — Two Days, Four Basis Points

I am not going to smooth this over, because the scorecard is worthless if I only report the parts that flatter me. The 10-Year did close above my line.July 23 at 4.71%, July 24 at 4.69%. That happened, it is in the Fed's own data, and anyone who wants to hold it against the call is entitled to.

Here is why I still call the level intact rather than broken, and you can judge the reasoning for yourself. A level breaks when the market accepts a new range above it. Four basis points for two sessions followed by an immediate rejection back under is not acceptance — it is a failed push. In my read, getting tagged twice and thrown back both times is evidence for resistance at that level, not against it.

Watch No. 3 called it “tested” on July 23. Looking back with the benefit of what happened next, that was the right word.

What the Fed Actually Did on Wednesday

The Federal Open Market Committee held the target range at 3.50%–3.75% — the fifth consecutive hold. The vote was 9–3.

Now here is where almost every take you read today is going to get it backwards, and I am not going to be polite about it. You will see “three officials dissented” next to a soft day for stocks and you will assume three people wanted rates cut. Dead wrong. All three wanted rates higher. Cleveland's Beth Hammack, Minneapolis's Neel Kashkari and Dallas's Lorie Logan each dissented because they preferred to raise the target range by a quarter point at this meeting. That is not a dovish crack in the committee. That is three people arguing the Fed is not tight enough.

Chair Kevin Warsh's language pointed the same way: “This Fed will not waver,” and a reminder that five-plus years of above-target inflation “cannot be cured in nine weeks.” The statement still describes inflation as elevated relative to the 2% goal and activity as expanding at a solid pace.

So understand what the 10-Year absorbed on Wednesday and still didn't break through:a hawkish hold, three officials pushing for a hike, a Fed chair promising not to waver, and crude up better than six percent. If there were ever a day built to blow through a ceiling, that was it. It didn't.

The Split Almost Nobody Will Report

Something didbreak on Wednesday. It just wasn't the 10-Year, and the difference matters enormously if you are financing a house.

Short end down, belly holding its line, long end to a nineteen-year high. Bond desks call that shape a steepener, and in plain English it says something specific: the market lowered its bet on an imminent hike while simultaneously demanding more compensation to lend for thirty years. That is not a market relieved by a hold. That is a market that has stopped worrying about the next meeting and started worrying about the next decade.

And notice which instrument is telling you that. Not the 10-Year — the 30-Year. I think that distinction is going to get flattened into “rates hit a 19-year high” in a lot of coverage today, and it is the wrong read for a mortgage borrower. Your 30-year fixed does not price off the 30-year Treasury bond. It prices off the 10-Year and mortgage-backed bonds, because most mortgages get paid off or refinanced long before thirty years is up. That is my opinion on the mechanics, plainly labeled — but it is why the ceiling holding is the more relevant fact for you than the long bond breaking out.

Before I Take Any Victory Lap

A scorecard that only accumulates wins is a sales brochure, so here is everything working against the call:

The Watch, Scored Honestly

No. 2 (Builders Just Blinked) was about builder incentives rather than a rate level, so there is no rate call in it to score.

The Real Lesson Isn't the Number. It's the Kill Switch.

Those two days above the line taught me something more useful than being right did, so let me hand it to you straight. When the 10-Year printed 4.71%, I had no pre-committed rule for what would count as my call being wrong.I hadn't written one. So I was left deciding, after the fact, whether four basis points for two sessions was a break or a test — and that is exactly the moment when a person's judgment is least trustworthy, including mine.

I still think “test” was correct, and the week since supports it. But I was reasoning toward a conclusion I already liked, with no rule stopping me. A call with no stated kill switch quietly turns into a hope, and a hope will happily survive evidence that should have ended it.

Here is the difference I have watched play out for thirty years. An amateur takes a bad call personally. They go quiet, they sulk, they spend days or months off the horse — and the market does not wait for them to feel better about it. A professional treats a near-miss as paid tuition. Take the lesson, pivot, and use the thing that rattled you as a flywheel — come back around faster and smarter than you were before it happened.

So here is the flywheel, spinning: from here on, every Watch entry that makes a call names the level or the condition that kills it, in advance, in the post. For this one: I will consider the 4.67% ceiling broken on two consecutive weekly closes above 4.75% — not an intraday spike, not a two-day poke. Written down before the fact, so neither of us has to take my word for it later.

So What Now? I'm Not Handing You a New Number.

The call I already made is still on the table and still being tested, so I don't need a new one. What I can tell you is what to actually watch, which is not the next Fed meeting.

Watch the 10-Year and watch the gap between the 10-Year and the 30-Year. If that gap keeps widening, the long end is telling you it doesn't believe inflation is beaten, and eventually that argument reaches the part of the curve your mortgage actually prices off. If the gap narrows, the ceiling story gets stronger. I think Wednesday demonstrated something most coverage will miss entirely: those two instruments can move in opposite directions on the same afternoon, and only one of them is yours.

Why Keep Score Out Loud

Because I said I would, and because a market read you can't audit is worth exactly nothing to you. On July 15 I planted a flag on a specific number knowing the market might walk straight over it. It went and stood on the flag for two days. I am telling you that in the same post where I tell you the level held, because you cannot have one without the other and still call it a scorecard.

A forecaster who only publishes the hits isn't a forecaster, he's a marketer. The Watch continues — and now it continues with a kill switch written down in advance.

Related: The 4.67% Ceiling · Lock Before the Fed · Highest in a Year · ARM vs Fixed · VA Loans · Physician Loans