The three leading Wall Street indices rose yesterday. On the same day gold rose, bitcoin rose, and oil stayed high.
It looks irrational. Risk assets and safe havens are supposed to move in opposite directions, and when everything rises together the impression is that something is mispriced.
The explanation is very simple, and it lies in one sentence spoken by a central bank governor. And this morning a data print arrived that pulls in exactly the opposite direction, so it is worth starting from the end: what drove yesterday is no longer the only thing on the table.
Errors or inaccuracies are possible. Spotted something that looks wrong? Write to me and I will correct it.
What Moved Yesterday
| Close | Change | |
|---|---|---|
| Dow Jones | 53,686.11 | +1.18% |
| S&P 500 | 7,749.81 | +1.06% |
| Nasdaq | 26,584.06 | +1.40% |
| Gold | about $4,478 | +2.06% |
| Bitcoin | $81,263.99 | +5.12% |
| Brent crude | $95.82 | |
| Two-year yield | 4.34% | -5 basis points |
| Ten-year yield | 4.77% | -2 basis points |
The last two rows are the explanation for every row above them - and chiefly the first of the two. The two-year yield is the one that prices near-term rate expectations, and it moved more than twice as far as the long yield did.
(Yield figures are from the US Treasury's daily yield curve. The closing price of US crude for that session was not verified against a primary source, and so does not appear in the table.)
The Sentence
Christopher Waller, a governor of the Federal Reserve, said yesterday that he sees encouraging signs of disinflation - noting that inflation over the past three months had fallen to 4.76% - and made clear that if the trend continues he would support holding rates unchanged at the September meeting.
And the implied probability of a rate rise this month fell from 63% to about 50%.
That is a 13 percentage point move in pricing, within a day, on the remarks of a single governor.
Why That Moves Everything at Once
Interest rates do not affect equities alone. They are the yardstick by which every future cash flow is discounted, so a change in rate expectations travels through every asset whose value sits in the future.
The Nasdaq rose more than the Dow - 1.40% against 1.18% - and that is consistent: technology shares are long-duration assets, and they benefit more from falling rates. And indeed the standout names that session were software - ServiceNow, Intuit, Adobe, Meta and Microsoft.
Gold and bitcoin move on the same logic, even though neither produces any cash flow at all. Both are assets without a running yield, so the cost of holding them is the interest given up. When expected rates fall, that cost falls, and demand for them rises.
In other words: four things did not happen here. One thing happened, and four markets repriced it.
And the One Asset That Explanation Does Not Cover
Oil.
It barely moved yesterday - but it arrives after a sharp surge. On US Energy Information Administration data, US crude rose from $84.57 on 28 August to $91.48 on 1 September - 8.2% across two sessions - and it has held around that level since. Brent closed yesterday at $95.82.
And what moved it is not rates but the Gulf: the US president said America had carried out a "very heavy attack" against Iran, and later that the renewed strikes would be short-lived.
So this market has two engines running at once - rate expectations lifting every long-duration asset, and a war premium holding oil high separately from them.
And that is also why the picture looks "abnormal": one looks at a screen where everything is green and assumes a single cause explains all of it. There are two.
And How Much Oil Actually Passes Through the Strait
That question was left open in the previous piece, and it is relevant precisely because oil is priced on it.
The US energy secretary said 17 million barrels passed through the Strait of Hormuz on Monday - in his words, the highest daily volume since the fighting began. Those are Department of Energy estimates rather than an independently verified figure.
| The measurement | Who |
|---|---|
| 17 million barrels on Monday | US Department of Energy |
| Estimated running flow: about 8 million barrels a day | market participants |
| Six commercial vessels on Wednesday, against 11 on Tuesday | shipping trackers |
| The typical average | close to 13 a day |
The official explanation for the gap is real: tankers switch off transponders to avoid being hit, so satellite tracking undercounts. On the other hand, an estimate from a department that is a party to the conflict, with no published methodology, carries a different problem.
(In early August the same energy secretary estimated an average of nearly 9 million barrels a day; market analysts estimated 4 to 6 at the time.)
And This Morning the Jobs Report Arrived
And here the piece changes direction, because a few hours after the close described above, the number this entire discussion has been waiting for was published.
| The August employment report | |
|---|---|
| Nonfarm payrolls added | 162,000 |
| Economists' forecast | about 53,000 |
| Unemployment rate | 4.1%, unchanged |
| Number of unemployed | 7.0 million |
| Average hourly earnings | $37.75, +0.3% on the month |
| Average hourly earnings over the year | +3.1% |
| Revision to June | from 20,000 to 31,000 |
| Revision to July | from minus 23,000 to plus 21,000 |
That is the strongest monthly gain since March. The leading sectors were food services and drinking places and local government education; the information industry lost jobs.
And the bottom two rows of that table are, in my view, the most important in the whole report - and they were barely discussed.
On 7 August the odds of a September rate rise fell sharply, and that happened on the back of the July employment report, which showed a loss of 23,000 jobs. That was the number that grounded the claim the labour market was cooling.
This morning that number was erased. July was revised from a loss of 23,000 to a gain of 21,000 - a difference of 44,000 jobs - and June was revised up by a further 11,000. So it is not only that August was strong: the weak month on which the cooling argument rested did not exist in the form it was reported.
(The implied probability of a rate rise moved back up after the release. I am not quoting a precise figure, because readings differ between pricing venues and they move while a session is open.)
The Disagreement That Has Opened Inside the Fed
And that makes the following picture particularly interesting, because it is no longer a disagreement between two people but between two people and a data print.
Seven days ago, at Jackson Hole, Fed Chair Kevin Warsh stressed that six-month inflation stood at 4.1% against 3.7% over twelve months - that is, the shorter rate running faster than the longer one - and said the summer readings "do not tell me that underlying trends have meaningfully improved".
Yesterday a governor of the same central bank pointed to inflation of 4.76% over three months and called it an encouraging sign of disinflation.
The two can be reconciled, and it is worth being fair about this. These are different measurement windows, and Waller is describing a direction - a rate that is falling - while the Chair described a level that remains high. Those are two different statements about the same data, not two contradictory facts.
But they do show the committee is not speaking with one voice, and that in itself is information. Warsh said in that same speech that he is "committed to a discipline, not to a decision", and that he is narrowing forward guidance.
And this is the practical consequence of that narrowing: when the Chair does not commit, every governor who speaks becomes a signal. Waller's remarks moved rate pricing yesterday more than the Chair's own opening address did - and this morning, a single employment print moved it back.
And One Detail Not to Miss
Broadcom's shares fell 2.74% yesterday to $357.16 - after reporting revenue growth of 86%, a 216% jump in net income, and guidance for 93% growth next quarter.
And during the session it fell far further: the intraday low was $342.33, that is 6.8% below the prior close, and the shares recovered a large part of the decline by the bell.
The explanation offered: the results were "not enough to keep investors happy", because the magnitude of the beat was not large enough relative to the company's AI exposure.
This is the same pattern seen this week at Palo Alto, which beat guidance for a fourth consecutive quarter and fell in three of them. The bar is not the guidance - the bar is whatever is already in the price, and that figure is published in no report.
הזווית שלי
דעה אישית של אילן אברמוב - לא ייעוץ ולא המלצה
What I take from these two days is mainly a correction to an instinct, and I think it is a common one.
When the whole screen is green - indices, gold, crypto and commodities - the impression is that something has come unmoored. Yesterday it looked that way and was not. A five basis point fall in the two-year yield, arising from a change in rate expectations, precisely explains why technology shares, gold and bitcoin all rose on the same day in the same direction. For this purpose they are not different assets - they are all assets whose value sits in the future, and that day the future became cheaper to discount.
What was genuinely unusual is oil, which rose for an entirely different reason through an entirely different mechanism. And that is exactly the detail that vanishes when one looks at "a green day".
And what troubles me more than the prices is how little it takes to move them. Thirteen percentage points of rate pricing moved yesterday on the words of a single governor - not on an inflation print, not on a committee decision, and not on the Chair's speech. And less than a day later, one employment print pulled that same pricing back. In a world where the central bank is deliberately narrowing forward guidance, that is not a malfunction - it is the designed outcome. The market is required to infer for itself, and it infers from whatever signal it can find.
And the thing I take most from all of it is the revision to July. A month reported as a loss of 23,000 jobs turned out to be a gain of 21,000. That is not a failure - it is how a data series that updates as information accumulates is built - but it does mean August's pricing rested on a number that no longer exists. Anyone who reads a first print as though it were final gets a sharper picture than the data justify.
And on 15 and 16 September, it is worth remembering that even after two days of all this movement, the implied probability sits around even. Which is the market saying, with a certain honesty, that it does not know - and I do not think anyone needs to pretend otherwise.
(It is important to stress: this is my personal opinion only, and nothing herein constitutes a recommendation to take any action.)
