From the archive · Wednesday, August 26, 2026
The mines came out of Hormuz and the flow did not come back — and while crude sold the geopolitics, gold and bitcoin priced the Treasury buying its own long bond.
They cleared the mines and nothing moved
Sponsored duration, unpriced chokepoint
Tuesday's settled session produced one headline and two unrelated trades. Washington announced that every mine had been detonated or removed from the international waters of the Strait of Hormuz, warned that any vessel laying more would be destroyed, and said the Space Force was watching the channel; separately, Iran and Oman reached a temporary reopening arrangement while negotiations continue. Brent fell 3.89% to $88.58 and WTI 3.12% to $82.36, Brent's largest single-session fall since 2026-08-04. The S&P 500 rose 0.32% to 7,677.28, the Nasdaq 0.66% to 26,151.30, and yields fell across the curve.
The reopening leg does not survive contact with the flow data. Transits through the strait were running at about five million barrels a day on Monday against more than twenty million before the war — roughly one in every five barrels consumed worldwide, now at a quarter of normal. Our own probability layer moved against the reopening on every independent measure it carries, on the same session: the odds of nought-to-twenty average daily transits on 31 August rose to 0.988, the highest reading of the window; the odds of even one thirty-ship day before month-end fell to 0.035, the lowest; and the odds of a negotiated Iran-Oman agreement by 30 September rose to 0.480. That last one is the informative move. Nobody bids a settlement for a waterway that has just been repaired. Tehran spent the same day threatening measures against forty-five vessels it says violated its transit rules, and promising retaliation against neighbours joining the sanctions campaign. Clearing the mines removed the instrument of coercion; it left the insurance, the war-risk premium and the rules of passage exactly where they were.
The larger leg had no headline attached to it. Gold closed at $4,694.50 and bitcoin at $78,487, both at three-month highs. Neither is an Iran trade — gold made its high on the day the war's central chokepoint was declared open. What they share is the US Treasury's announcement that it would double its purchases of its own long-dated debt. Gold is on course for its strongest month since September 1999 and bitcoin ran twenty percent in three days, its best since 2023, with $1.92bn of spot fund inflows last week. Those are not disinflation trades, and the inflation data does not support a disinflation reading either: core PCE is forecast at 3.3% year on year, the headline at 3.7%, and household one-year expectations at 4.3%.
The curve is the discriminator, and it takes a week rather than a day to read. On Tuesday the move looked parallel: five-year 6 basis points lower, ten-year 7, thirty-year 6. Across five sessions it is not parallel at all — the thirty-year has given 14 basis points to 5.17%, the ten-year 8, and the five-year 3. A disinflation impulse from a cheaper barrel reaches the front and the belly first, because that is where the policy path is priced. This one arrived almost entirely in the maturity whose issuer has said it will buy, and it arrived as a squeeze in long bonds rather than as demand from savers. The issuer has become the bid for its own long bond, and the assets that price sovereign promises are re-rating around it.
Two secondary readings are worth separating from the noise. First, the benign alternative — that crude fell because growth is cooling — is closed off by copper, which rose 1.65% to $6.714, within 0.20% of its highest close in two hundred and fifty-two sessions, with the futures crowd's net long at 78,648 contracts against 69,008 in early July. A demand scare does not do that. Second, the consumer is not slowing so much as splitting: confidence fell to its lowest since January while Lego posted record first-half revenue at both premium and value price points and Dick's cut guidance on Foot Locker comparable sales of -3.6%. Records at both ends of the price ladder with the middle cracking is a distribution problem, and an aggregate confidence number cannot see it.
This is the July credibility call maturing in a shape we did not forecast. We argued the institutional discount would show up as a steep long end; instead the steepness has been bought in, and the discount has migrated to assets no issuer can print. Yesterday's brief argued that coercion had bought a shorter wait rather than a reopening, and expected the constraint to be monetised as a transit fee. Tuesday marks that both ways: every fee horizon gave back — the August band to 0.035, September to 0.150, October to 0.305 — while the reopening leg was confirmed harder than we put it. The constraint is real; it is being taken as duration rather than as a toll. Three tests arrive in three days: the inflation print today, the year's largest chip result tonight, and a central-bank address on Friday whose silence on the balance sheet would be its loudest content.
Methodology. Gold at $4,694.50 is a three-month high, not a record: it is 12.91% below its 1 March close of $5,390.20, and silver is 40.54% below its 26 January peak. Probability bands are rebuilt from the current-probability time series rather than the vendor's prior-value field, which oscillates and is not a reliable prior day. Positioning figures are the maximum-exposure contract market for each commodity and describe one reporting category, not a census of buyers; the latest reporting date is 2026-08-18. Percentage moves are computed at build time from settled closes.