The SPR Hits 293.4mb (19 August 2026): A 4.4mb Crude Build That Was Really a 0.9mb Draw — and 172 Million Barrels That Have to Come Back With a Premium

The Strategic Petroleum Reserve fell another 5.3 million barrels to 293.4 million in the week to 14 August — the lowest weekly reading since 31 December 1982 — and the 4.4 million-barrel commercial "build" reported alongside it was smaller than the reserve draw that helped create it, so total US crude on the ground fell 0.9 million barrels. Every barrel of that build was on the Gulf Coast, where the reserve’s caverns are; Cushing, the WTI delivery point, fell to 21.3 million. And the barrels are a loan rather than a sale: the 172 million-barrel programme is an exchange, and the Department of Energy has been awarding tranches at return premiums of 26–28% in kind, which turns the eventual refill into contractual demand instead of a budget decision. Brent traded at $93.60 on Wednesday morning, 41.4% above a year ago; the loonie ended 0.10% weaker on the week.

Five months into this episode, the story keeps arriving through a single number that does not mean what it appears to mean. For a fortnight it was supply: which lane runs north, whose fee, whether an underwriter would cover it. On 12 August it became demand, when the IEA, OPEC and the EIA published within hours of each other and all three pointed the same way. This week it is neither. It is the buffer that has been quietly standing in for the missing barrels — how much of it is left, where the barrels land when they leave, and the fact that they have been lent rather than sold. The earlier sequence, which is what built the premium now being tested, is set out in full below and is unchanged.

On 2 August, seven OPEC+ producers signed off the final 188,000 barrels a day for September and a pause in quota increases for the rest of 2026, closing a month in which Brent gained 24%. That decision is unchanged and is set out in full below. What has changed since is the category of the obstacle, three times over. Through 3–7 August the dispute was maritime: which lane runs north, who supervises inspections, whether a fee is 3% or 7% — and then whether a fee-bearing deal is insurable at all, after a Lloyd's Market Association clause terminated hull cover for any vessel that pays to transit. Since 8 August the dispute is about sanctions, frozen assets and reparations, none of which the Iran–Oman channel can settle. Not one additional barrel has been restored across the whole sequence. What has changed is the traffic count, and it has gone the wrong way: 15 vessels on Friday, 11 on Saturday, 6 on Sunday, against roughly 130 a day before the conflict. And the currency at the end of the chain did nothing again — USD/CAD's official Bank of Canada rate printed 1.3942 on Monday against 1.3943 on Friday, on a session Brent rose 5%.

What actually happened on 19 August: a headline build that was really a draw, and a reserve at 293.4 million

The EIA's weekly petroleum status report, released Wednesday 19 August for the week ending 14 August, carried a bearish-looking headline and the opposite arithmetic underneath it. Commercial crude stocks rose 4.4 million barrels to 428.8 million. In the same week the Strategic Petroleum Reserve fell 5.3 million barrels to 293.4 million. Because reserve barrels are sold or exchanged into commercial tanks rather than destroyed when they leave the salt caverns, the second number is partly the source of the first — and this week it is the larger of the two. Total US crude on the ground, commercial plus reserve, fell 0.9 million barrels to 722.2 million.

That is the same accounting distinction this post applied to the 17.4 million-barrel build a week earlier, with one difference that matters: last week netting the transfer shrank a build, and this week it erases it. A reader working only from the headline saw American crude inventories grow for a second straight week during a closed-strait supply shock. What the full table says is that the country held less crude on 14 August than it did on 7 August, and that the commercial "surplus" was assembled by moving the emergency buffer into the account the weekly headline measures.

The level is the other finding, and it is more precise than the round number suggests. The EIA's weekly reserve series begins on 20 August 1982 at 270.5 million barrels, on the way up. The last week it printed below 293.4 million was 31 December 1982, at 293.2 million; by 7 January 1983 it was 294.8 million and rising every week thereafter. So this is not merely a 1983 low. It is the lowest weekly reading in the entire published history bar the reserve's first four months of existence, and the only other time the number was this small the direction of travel was the reverse of today's.

Every barrel of the build was on the Gulf Coast

The regional breakdown removes any remaining ambiguity about where the crude came from. Of the 4.4 million-barrel national build, the Gulf Coast (PADD 3) accounted for 8.4 million — more than the national total, because every other region drew down.

The Gulf Coast is where the reserve is. All four Strategic Petroleum Reserve sites — Bryan Mound and Big Hill in Texas, West Hackberry and Bayou Choctaw in Louisiana — sit on the Texas and Louisiana coast, and their barrels are delivered into the pipeline and terminal system of PADD 3. A national build that is entirely one district, and specifically the district containing the caverns that lost 5.3 million barrels in the same week, is not a story about slack demand. It is a story about geography.

Cushing, Oklahoma is the detail worth holding onto. Cushing is the delivery point for the WTI futures contract, which makes its tank level the physical anchor of the US benchmark price. It fell 1.3 million barrels on the week to 21.3 million, against 23.5 million a year ago and 28.2 million two years ago. Operators generally treat something in the region of 20 million barrels as the practical floor — below that, tanks hold volume that cannot be pumped out cleanly, and the hub loses its ability to absorb or supply at short notice. The benchmark's delivery point is tightening while the national headline says the opposite, which is a reasonable description of why the flat price has not behaved the way a two-week run of builds would imply.

The barrels are a loan, and the premium is the point

Here is the part of the reserve story that the weekly inventory number cannot show, and it changes what the drawdown means. The 2026 release is not a sale. It is an exchange: companies borrow crude from the reserve and contract to return the original volume plus additional barrels at a later date. The EIA states the structure plainly — the release "requires the original volume of oil, plus additional barrels, to be returned to the SPR at a later date."

The scale is documented in the Department of Energy's own award notices. The programme is 172 million barrels, described by DOE as the US portion of an IEA-coordinated collective action of 400 million barrels. By its 10 June RFP, DOE said it had awarded 133 million barrels across four solicitations and was tendering a further 40 million. The return premiums are the striking figures: DOE cited a 26 percent premium on earlier exchanges, and on a 53.3 million-barrel tranche awarded on 11 May it reported a premium of approximately 28 percent — 15.1 million barrels above the volume lent.

"This exchange will help move oil swiftly to refiners, ease short-term supply pressures, and ensure the Strategic Petroleum Reserve continues to grow stronger through the return of premium barrels," DOE Acting Assistant Secretary Curt Coccodrilli said in the June notice.

The mechanism this sets up is worth being careful about, because it is easy to overstate in either direction. What an exchange does is convert a one-off supply addition into a later obligation to buy. When a government sells reserve crude, the refill is a budget decision that can be deferred indefinitely; when it lends reserve crude at a premium, the barrels come back because a counterparty signed for them. That makes the eventual return leg a firmer source of demand than a refill programme — and a larger one, since the premium means more crude goes back in than came out. What it does not do is fix a date, a price or a rate. The published notices set out volumes and premiums; they do not commit anyone to a delivery schedule the market can trade against, and DOE's April award language put the first tranche's return at "by next year" rather than on a stated day.

The arithmetic on what is left is straightforward from the weekly series. The reserve held 415.4 million barrels on 20 March, before the drawdown began. At 293.4 million it has released 122.0 million — about 902,000 barrels a day measured from its 3 April level of 413.3 million, a rate of unscheduled supply roughly comparable to a mid-sized OPEC member's output, arriving with no quota and no announcement. Against a 172 million-barrel authorisation, roughly 50 million barrels remain to be delivered. If the programme runs to its stated size, the reserve troughs somewhere around 243 million barrels — and then has to start taking barrels back.

The demand ceiling is now showing up in volumes, not just forecasts

A week ago the demand side of this shock was a pair of forecast revisions. This report puts some of it into observed data. Refiners ran at 97.2% of operable capacity, with crude inputs of 17.4 million barrels a day, up 215,000 on the week — flat-out operation, which is what record cracks pay for. Domestic production printed 13.83 million barrels a day, up 448,000 on a year earlier. Yet the four-week measure of total products supplied — the EIA's proxy for US consumption — came in at 20.5 million barrels a day, down 2.9% year on year, against −2.1% in the previous report. Jet fuel supplied swung from +3.8% year on year to −6.3% in a single report, and distillate from +1.9% to −0.8%.

Two cautions on reading that table. Four-week averages can move this much on one soft week plus a revision, so a single report is a data point rather than a trend, and jet fuel in particular is a volatile series on a small base. But the direction is consistent with what both agencies wrote down on 12 August for a stated reason — that fuel prices are high enough to remove consumption — and the composition fits the mechanism rather than cutting against it: the fuels whose cracks went to records are the fuels whose volumes are now falling fastest.

Meanwhile distillate stocks fell another 1.5 million barrels to 105.6 million, about 13% below the five-year average and 9.0% below a year ago, with refiners running at 97.2%. That combination — maximum runs, still drawing — is the inventory signature behind the record diesel cracks, and it is the reason a crude build and a tight product market can coexist without either number being wrong.

The price went up anyway, and the loonie did not

Brent traded at $93.60 a barrel at 6:30 a.m. ET on Wednesday 19 August, up $1.18 or 1.27% on the day, per Fortune — a fourth consecutive session of gains, 5.71% higher than a month earlier and 41.4% above its level a year ago. Two weekly builds in a row did not stop it, which is what the netted table predicts and the headline does not.

The currency at the end of the chain repeated its performance for the month. The Bank of Canada's official noon rate for USD/CAD printed 1.3889 on 18 August against 1.3875 on 14 August — the Canadian dollar 0.10% weaker across a stretch in which Brent rose roughly 5.7%. Between them sat 1.3865 on 17 August, a 0.07% CAD gain. For a currency whose commodities factor is supposed to run through crude, four months of this episode have produced a consistent answer: Canada sells a discounted heavy grade under long-term contract into one customer, so the marginal barrel that sets Brent is not the marginal barrel that sets Canada's terms of trade, and the interest-rate factor has had more leverage on the pair all month than the commodity has. The same disconnect is documented across the whole Hormuz episode and, in the opposite direction, in gold against real yields.

What would change the picture

Four observables, in order of how much they would move the analysis. First, a commercial build that repeats with no reserve contribution and no import surge — imports actually fell 746,000 barrels a day this week, so the trade-flow explanation for the previous build has already weakened, which makes the transfer the load-bearing part of this one. Second, the reserve draw stopping: roughly 902,000 barrels a day of unscheduled supply leaving the market is a supply event that arrives with no notice, and on the programme's own stated size there are only about 50 million barrels left to give. Third, the return leg starting — the first tranche of premium barrels flowing back would be the first identifiable new source of crude demand in this episode, and unlike a refill it is contractual. Fourth, and still the only evidence that cannot be renegotiated, voided by an underwriter or attached to a sanctions demand: the number of vessels the strait actually passes.

The week before, 12–14 August: a reserve below 300 million barrels, two demand cuts, and a 24-hour round trip

Three releases landed on Wednesday 12 August — the IEA's monthly Oil Market Report, OPEC's, and the EIA's weekly petroleum status data — and all three pointed at demand. The price took one session to agree and one session to change its mind.

The build that was partly the reserve changing pockets

The EIA's weekly series put US commercial crude stocks at 424.4 million barrels for the week ending 7 August, up 17.4 million on the week — the largest weekly build since January 2023, against a consensus that had expected a draw of about 1.4 million. Reuters reported crude imports rose 1.14 million barrels a day week-on-week while exports fell 627,000 a day. That is a trade-flow build: barrels that were already at sea arriving into US tanks, and barrels that would have left staying put.

There is a second contributor, and it is the one that changes what the number means. In the same week the Strategic Petroleum Reserve fell from 304.8 million barrels to 298.7 million — a draw of 6.1 million. SPR barrels do not evaporate when they leave the salt caverns; they are sold or exchanged to refiners and terminals, and they land in commercial inventory. So roughly a third of the headline build is the same crude counted in a different pocket.

The level is the second finding. 298.7 million barrels is the first sub-300-million reading in the entire weekly history since 28 January 1983, when the reserve stood at 298.4 million and was being filled rather than drained. The pace is more informative than the level: the same series read 413.3 million on 3 April 2026, so 114.6 million barrels have left in 126 days — an average of about 910,000 barrels a day of unscheduled supply that has been quietly absorbing part of the Hormuz shortfall for four months. Nothing announces when that stops. It is not a quota, it has no published schedule, and it is the reason the market has been able to treat a closed strait as a price event rather than a rationing event.

Both agencies cut 2026 demand — and both named the same cause

The IEA's August report, published 12 August, now expects world oil demand to decline by 1.6 mb/d in 2026, which it describes as 510 kb/d more than its July estimate. Supply is forecast to fall 4.3 mb/d to 102 mb/d, with a rebound of 8.3 mb/d to 110.3 mb/d in 2027. The agency put total observed oil stocks at just below 7.9 billion barrels, down 410 million since the war began — an average draw of 2.7 mb/d. Its stated reason for the demand cut is direct: elevated fuel prices "are putting further downward pressure on oil use," while the closure "disrupts international supply chains and curtails product availability."

OPEC, reporting the same day, still sees growth rather than contraction, but cut it for the fourth consecutive month. Reuters put the new 2026 figure at 580,000 barrels a day, against 800,000 in the July report; in level terms the 2026 forecast moved to 105.74 mb/d from 105.94 mb/d. OPEC raised its 2027 growth forecast to roughly 2.2 mb/d.

The two agencies disagree about the sign and agree about the direction of travel, which is the useful part. Neither revision is a statement that the supply shock is over.

Why a supply shock builds its own ceiling

This is the mechanism the week actually delivered, and it runs in one direction only. A closed strait removes crude. Scarce crude raises the price of the products refined from it faster than it raises crude itself, because refining capacity cannot be relocated to where the barrels still are — the IEA recorded Atlantic Basin refining margins at all-time highs in July as diesel, jet fuel and gasoline cracks surged, with diesel exports from Russia, the Middle East and Asia running 1.3 mb/d below a year earlier and jet fuel exports about 670 kb/d lower. Record cracks are pump prices. Pump prices are the thing that changes behaviour. And the consumption that behaviour removes is exactly what both agencies wrote down on 12 August.

So the price does have a ceiling, and the ceiling is endogenous — it builds itself out of demand that has been priced out. What it is not is a supply recovery. A barrel that is no longer consumed because diesel is unaffordable is not the same asset as a barrel that has come back through Hormuz, and the two have opposite implications for what happens if the strait reopens. Conflating them is the single most common error available on a week like this one.

The 24-hour round trip

Thursday 13 August was the session that priced the demand read. Brent settled down $1.91 at $87.07, snapping a six-session winning streak; WTI fell $2.02 to $81.25, ending a five-session run. Both benchmarks lost more than 2%.

It lasted a day. On Friday 14 August Brent rose 1.7% to settle at $88.52 and WTI gained 1.4% to $82.40, after US Defense Secretary Pete Hegseth told reporters that US forces could maintain the blockade of Iranian ports indefinitely, and Treasury Secretary Scott Bessent said in a Newsmax interview that measures aimed at the "economic isolation" of Iran would be of a kind that "have never been seen." Per CNBC, both benchmarks finished the week more than 5% higher.

Running underneath both sessions: Bloomberg reported that the Houthis struck Saudi Aramco's Jizan refinery with drones on 13 August, the second attack on that facility in under a week. The detail that matters is one this post could not report on 10 August: the 400,000 bpd plant was already out of service after a late-July attack damaged its integrated gasification combined-cycle complex and tank farm, with the restart pushed back to around the end of the month. The Red Sea end of the bypass route is not a contingency being tested — part of it is down.

Two days, one message. A demand revision is a forecast; a blockade statement is a constraint on barrels that exist now. When the two arrive within 24 hours of each other, the constraint wins, because the premium in the price is compensation for duration and the blockade language lengthened duration while the demand cuts did not shorten it.

The loonie, again, moved on the dollar

The Bank of Canada's official rate tells the same story it has told all month. On Thursday, the day Brent fell 2.1%, USD/CAD went from 1.3930 to 1.3938 — a move of 0.06% against its principal export dropping nearly $2 a barrel. On Friday, when Brent rose 1.7%, USD/CAD fell to 1.3875, a 0.45% CAD gain, on a session whose driver was US rate expectations rather than crude. Across the week, 1.3943 (7 August) to 1.3875 (14 August) is a 0.49% CAD gain — a third consecutive weekly decline in USD/CAD, achieved on a week in which oil rose more than 5% and the loonie's own biggest single day came from the dollar side of the pair.

For a currency the commodities factor is supposed to score through crude, that is a persistent and instructive failure of the textbook channel — the same one documented for the whole Hormuz episode, and the mirror image of what happened to gold when real yields rose. Canada exports a discounted heavy grade under long-term contract into a single customer; the marginal barrel that sets Brent is not the marginal barrel that sets Canada's terms of trade, and the rate factor moving in the same week has more leverage on the pair than the commodity does.

What actually happened on 10 August

The framework arrived on Friday. By Monday it had been replaced by a different kind of document, and the price paid for the difference.

On Saturday 8 August, Mohammad Bagher Zolghadr, secretary of Iran's Supreme National Security Council, said the Strait of Hormuz would remain closed until Washington "corrects its behaviour," and set out six conditions: an end to US threats against Iran and to insults to what he described as the country's national and religious values; a permanent end to attacks against Iran and its allies in Lebanon, Palestine, Yemen and Iraq; the lifting of the US naval blockade and the withdrawal of US naval and air forces from around Iran; compensation for damage from two "imposed wars"; the lifting of sanctions; and the unconditional release of frozen Iranian assets. "Whenever the United States accepts Iran's conditions, the Strait of Hormuz will certainly be reopened," he said. President Masoud Pezeshkian, speaking separately, said "now is the best time for an agreement" and expressed hope Iran could move beyond "neither war nor peace" — so the two statements are not identical in tone, and the market had to price both.

Brent's October contract settled about 5% higher at $87.72 on Monday 10 August. WTI settled about 5% higher at $82.13. It was a fourth consecutive session of gains, a 6.8% recovery from Friday's $82.12, and it retraced essentially the whole of the 3–4 August slide — the October contract had been trading near $87.93 before the selling began on 2 August. Roughly two weeks of repricing, in both directions, has ended almost exactly where it started.

Now the part worth being precise about, because it is the mechanism rather than the headline.

The US response the same weekend moved in the opposite direction and made the gap symmetric rather than one-sided. President Trump told Axios on Sunday that the United States was "only semi-negotiating" with Iran, indicating he would rely on the naval blockade rather than a further wave of airstrikes: "We are just watching Iran with its huge inflation and the fact they have no money." He had cancelled a planned attack on 1 August to allow for negotiations. On Monday he addressed the compensation demand directly — with a counter-demand. "We're going to ask for money for the damage they've done over a 50-year period," he said, adding that "the only one that has control of the Strait of Hormuz right now is the United States Navy."

That reciprocity is the genuinely new feature. A dispute in which one party wants compensation can be settled by paying or refusing. A dispute in which both parties demand compensation from each other has no obvious mechanism at all, and it is not a question a lane diagram can resolve. Iran's foreign ministry spokesman Esmaeil Baghaei restated the binding item on Monday, according to the state news agency Tasnim: "As long as the U.S. naval blockade continues, the necessary conditions for the reopening of the Strait of Hormuz do not exist." Qatar's foreign ministry, meanwhile, described the Oman–Iran negotiations as at an advanced stage — which is the same split between channels that has run through this entire episode, and the reason the price keeps reversing on documents.

The traffic count went the wrong way

This post has argued from the start that the only evidence in this story that cannot be renegotiated is the number of ships. That series has now been updated, and it did not improve.

Al Jazeera reported transits of 15 vessels on Friday, 11 on Saturday and 6 on Sunday, against roughly 130 a day before the conflict began on 28 February. US Central Command has redirected 55 commercial vessels since reinstating the blockade in July, and the blockade has brought Iranian crude exports from Kharg Island to a complete halt. Al Jazeera also reported at least 64 violent incidents and 17 deaths involving commercial vessels in the region since the war began. A tanker operated by Abu Dhabi National Oil Company was attacked in the strait over the weekend.

Set that against a fortnight of diplomacy — coordinates, a draft text, a finalised framework, a list of conditions — and the physical series has gone 8, 15, 11, 6. It is not a recovery interrupted; it is a low count that got lower while every document suggested the opposite direction. For a factor framework this is the cleanest available illustration of why the commodities input and the risk-sentiment input are scored separately: the price has round-tripped roughly 10% down and then 10% back up over six sessions, while the underlying constraint has done nothing but tighten slightly.

The buffer behind the price is nearly gone

One number changes the character of everything above, and it has nothing to do with Iran.

The EIA's weekly series put the US Strategic Petroleum Reserve at 304.8 million barrels for the week ending 31 July 2026. That is the lowest reading since February 1983 — in the entire weekly history back to 1982, the last comparable level is 303.7 million barrels on 18 February 1983. Wire reporting on 10 August put the current level below 300 million; the next official EIA release is 12 August, so treat the sub-300 figure as reported rather than published.

The trajectory is more informative than the level:

Roughly 108 million barrels have left the reserve in four months, an average of about 900,000 barrels a day of supply that has been quietly offsetting part of the Hormuz shortfall — nearly five times the size of the September OPEC+ quota increment, and invisible in any production number. Two things follow. First, part of the reason the price has spent the summer trading documents rather than scarcity is that a non-market supply source has been filling the gap. Second, that source is now four-fifths depleted relative to April, and the drawdown rate has itself been slowing.

Put it alongside the OPEC+ decision set out below and the position is unusual in a specific way: the alliance's pre-committed supply runs out with September, and the largest strategic reserve in the world is at a 43-year low, at the same time. Neither fact moves today's price. Both widen the distribution around every subsequent headline — which is a risk-sentiment statement, not a commodities one, and it is the reason a currency framework can register a deteriorating backdrop on a day the oil price does nothing.

The bypass was struck at the refinery, not the tanker

The section further down this page argued that the only meaningful Hormuz bypass terminates on the Red Sea, and that this makes the two chokepoints one risk rather than two. On Sunday 9 August that argument stopped being an inference.

Houthi forces attacked Saudi Aramco's refinery at Jizan on the Red Sea coast — a facility that processes about 400,000 barrels of crude a day — using a drone, according to military spokesman Yahya Saree. Saudi Arabia's energy ministry said a fire broke out and was later extinguished, with no injuries. The same day, the Yemeni Red Sea port of Mokha, close to the Bab al-Mandeb strait, was attacked, killing at least seven people. The refinery strike came two days after Saudi Arabia signed the Mecca Joint Defence Pact with Turkey and Pakistan.

The escalation from tankers to fixed infrastructure matters for pricing in a way the incident report does not capture. A tanker attack raises a war-risk premium on a voyage; the vessel usually completes it. A refinery is a fixed asset with a known address and a published throughput, and it cannot be rerouted, insured around or convoyed. What the Jizan strike demonstrates is that the Red Sea leg of the workaround is now exposed at both ends — the water and the plant — which is the same structural point this page has been making about resilience arithmetic, now with a physical example attached.

And it landed two days before CPI

The one channel that did transmit on Monday was the rate channel, in the United States rather than Canada.

US Treasury par yields rose across the curve: the 10-year went to 4.72% from 4.65% on 7 August, the 2-year to 4.25% from 4.19%, and the 30-year to 5.25% from 5.19%, per the Treasury's daily yield curve. A 7 basis point move is not dramatic in isolation. Its timing is the point: it arrived on a 5% oil day, two days before the July CPI release on 12 August, and the direction says the bond market read the rally as an inflation impulse rather than a growth shock. Had it read it as a growth shock, yields would have fallen.

That is the transmission path this site keeps separate from the commodities factor and it is worth naming plainly. A supply-led oil rally reaches the dollar through the rate factor — energy feeds headline inflation, headline inflation feeds policy expectations, policy expectations feed the currency — while it reaches the Canadian dollar through the commodities factor, which is currently crowded. Same shock, two different channels, two different outcomes: the dollar's rate input moved and the loonie did not. What July CPI actually contains, and why the Cleveland Fed's own nowcast disagrees with the pump-price narrative, is set out in the CPI preview.

What happened on 7 August

The framework arrived, and the price ignored it.

Hassan Ghashghavi, spokesman for the Iranian parliament's national security commission, told the state news agency IRNA on Friday that the "general framework" of the agreement with Oman had been finalised and that "the final text and details will be announced soon." He added the qualification that matters: implementation is "subject to final approval at the highest decision-making levels." Iranian officials briefing state media had already made a second condition explicit — that any full reopening of the strait depends on the United States lifting its own naval blockade of Iranian ports.

Brent eased to $82.12 a barrel, down 0.45%. WTI eased to $77.06, down 0.30%. On Monday and Tuesday, the mere prospect of this announcement had taken roughly 10% out of Brent. By Friday, its arrival was worth 45 basis points.

That collapse in sensitivity is the week's real finding, and it is not explained by scepticism about whether a deal will be signed. It is explained by something that was published two weeks before any of this began, in London, by an insurance body.

The sanctions half of that exposure has a name and a date. The United States has designated Iran's Persian Gulf Strait Authority, the entity Tehran established in May 2026 to operate the waterway. It is the body that would collect a transit fee. So the counterparty on the receipt is the reason the receipt is a liability, and the problem does not dissolve if the percentage falls: 3% paid to a designated entity carries the same character as 7%.

This is why the specific numbers under negotiation, reported by Reuters on 5 August, matter less than they appear to. Iran was seeking 5% to 7% of the value of cargoes; Oman was discussing around 3%; Washington wanted none. Reuters, citing four industry sources, reported the resulting arrangement was not feasible for the shipping industry, with shipping groups warning that compulsory charges would be "a toll in all but name" and could undermine the international rules governing transit through straits. Gulf negotiators have pressed for regional supervision of ship inspections and for any fee to be voluntary — and voluntary is the load-bearing word in the whole dispute. A genuinely voluntary fee is compatible with the Lloyd's clause only if a vessel that declines to pay still transits. If refusal means no passage, the fee is compulsory in substance whatever it is called in the text, and the insurance consequence attaches anyway.

The same reporting carried the concession that would ordinarily have been the headline. The proposed deal would give Tehran control over ships entering the Gulf — among the largest concessions to Iran of the conflict — and one source told Reuters that "the concession has already been made regarding some form of control over Hormuz." How "control" is to be defined is unresolved. A market that had sold 10% on the prospect of a deal, and bought 3.8% back on the prospect of a bad one, responded to confirmation that the central concession had been granted by moving less than half a percent.

One further piece of Friday's tape belongs in the risk column rather than the diplomacy column. Saudi Arabia, Turkey and Pakistan signed the Mecca Joint Defence Pact, under which "any armed attack against any one of the three states shall be regarded as an attack against them all." A mutual-defence commitment spanning the largest Gulf exporter, a NATO member and a nuclear-armed state does not change any barrel's routing this month. What it changes is the width of the tail: it raises the number of parties who are formally committed to respond if the conflict widens to Saudi infrastructure — which, as set out further down this page, is exactly where the only meaningful Hormuz bypass terminates.

What actually happened on 6 August

For three sessions the market traded the idea of a deal. On Thursday it got its first look at what one side thinks the deal says, and reversed.

Iran's semi-official Fars news agency published the initial text of a strategic plan for managing the Strait of Hormuz, citing a member of Iran's parliament. Under that draft, passage would be prohibited for vessels belonging to the United States, Israel and other countries Iran regards as hostile. Israeli-linked cargo, whether military or civilian, and vessels involved in actions against Iran and its allies would also be barred. Ships linked to countries and individuals judged to have caused damage to Iran could not transit until that damage was compensated. Violators would face penalties of up to 20% of the value of the cargo aboard. The plan sits with the Iranian parliament's National Security Commission, which has invited outside specialists to submit recommendations before it is finalised.

Crude repriced immediately. Brent closed up 3.8% at $82.49 and WTI settled about 2.8% higher at $77.29. The US response arrived the same day and conceded nothing: "Any temporary routes will be without any impediments — meaning no approvals or permissions and no tolls or charges," a US official told CNBC, adding that "the Strait of Hormuz is an international waterway and no party controls the lanes or the ability to transit through them."

What makes the day genuinely hard to read is that two incompatible Iranian accounts were in circulation at once. Separately from the Fars draft, an Iranian government official connected to the talks said shipping would face "no fees or tolls" under the temporary Iran–Oman arrangement, with inbound traffic through Iranian territorial waters and outbound traffic on a route closer to Oman, and said the deal was designed to restart a 60-day negotiating cycle tied to the US–Iran memorandum of understanding signed on 17 June. Under that June memorandum Tehran was to allow commercial vessels through the strait free of charge for 60 days, and Washington was to lift its naval blockade of Iranian ships, among other commitments. The same official said the International Maritime Organization and the United States would take part in the announcement, and a Gulf government official said the Gulf Cooperation Council had endorsed the deal. Deputy Foreign Minister Kazem Gharibabadi, speaking to state broadcaster IRNA, said the US was ready to "return to commitments" while insisting that "the path of understanding is between Iran and Oman, and no negotiations with the US have taken place during this period"; in a Thursday statement he said the Iran–Oman understanding was close to being finalised and that Iran had demanded the routes change because the "old routes no longer meet the current difficult conditions that have jeopardized the national security of the Islamic Republic of Iran."

A parliamentary draft is not a signed agreement, and a bill under expert review is several steps from law. But a price does not wait for ratification. What the market bought on Thursday was not the draft becoming binding — it was the information the draft carried about the distance still to travel.

How the week got there: 2–5 August

The sequence is short and the order of it is the whole point.

On Sunday 2 August, President Trump announced he was halting a threatened major attack on Iran, saying the US had "just been asked by Iran, and other Middle Eastern Countries, to hold off any attack in that the perimeters of a deal has been agreed to." He described that deal as including the "Immediate, Complete, and Total OPENING OF THE HORMUZ STRAIT, and an end to Iran's nuclear threat," and said talks would resume on Monday, covering the strait before moving to the nuclear programme. He also said the US would "hold it" and watch how negotiations developed, while remaining ready to act "at any time."

Crude did not wait for the Monday session. Reuters reported Brent down $5.52, or 6.28%, at $82.41 a barrel as of 2202 GMT on 2 August, with WTI off $5.27, or 6.22%, at $79.40. By 03:27 GMT on 3 August Brent's October contract had recovered part of that to around $83.51, down about 5.1%, after trading more than 7% lower earlier; WTI was near $79.87, down 5.67%.

Now the part that did not happen. The Strait of Hormuz did not open. On 3 August it remained largely closed, with tankers still facing attacks and forced turnarounds and no agreement concluded. Vessel-tracking showed very little traffic through the strait early on Monday, though some vessels transit without broadcasting a position. An empty tanker, the Velos Amber, moving through the Omani lane appeared to be subject to an Iranian warning action on the Sunday; it continued and was off the UAE coast by Monday morning. Al Jazeera reported on 2 August that ships were stuck in the northern corridor in Iranian waters, that passage was possible only if Iran's armed forces allowed it, and that no breakthrough had been reached on managing traffic. Iran's acting defence minister, Brigadier General Majid Ebn-e-Reza, described the pullback as "within the context of psychological operations and a war of calculations."

Nor do the two governments agree on what is being negotiated. Iran's foreign ministry stressed that Iran is not negotiating with the United States; spokesperson Esmaeil Baghaei said Iran was in talks with Oman about establishing a safe, "temporary" route for vessels. Indirect contact through mediators is not the same thing as a bilateral negotiation, and both statements can be true at once — but the distance between them is the risk in the price.

Tuesday 4 August did the same thing again, harder. Brent's October contract settled at $79.36, down 5.3% and its lowest close in three weeks, with the previous low on 10 July; WTI's September contract settled at $75.77, down 5.7%. Compounding that with Monday's 5.1% puts roughly 10% of Brent's value in the hands of two sessions. Qatar said an interim proposal had been prepared, and both Washington and Tehran indicated that talks to restore access were progressing. Rebecca Babin, a senior energy trader at CIBC Private Wealth Group, named the mechanism as plainly as anyone has: "This is a market that consistently reprices risk on the prospect of flows resuming rather than the details required to achieve them."

Wednesday 5 August did something more interesting than a third leg down. Brent's October contract settled at $79.45, up 7 cents — effectively unchanged — while WTI's September contract settled at $75.22, its lowest in almost a month. The flat close is the finding, because the session contained two large and opposing pieces of news. Iran said it had agreed the coordinates of a Hormuz route with Oman, which is the most concrete step yet. And the Houthis claimed missile attacks on two Saudi oil tankers. A market that had sold 10% in two days on the prospect of a deal declined to sell any further on confirmation that the deal's geography exists.

Treasury Secretary Scott Bessent said an agreement could come "today or tomorrow." President Trump said "We'll know in 48 hours," adding that "The strait is going to be open very soon, or they're going to get hit very hard," and on the question of transit fees, "I'm not going to let them charge." Secretary of State Marco Rubio put it more carefully: "there's been progress made in those talks, but not finality yet." Iran's foreign ministry spokesman Esmaeil Baghaei described talks aimed at lanes that "uphold sovereign rights while also addressing the national security considerations of both Iran and Oman," and said differences remained over shipping operations. Iranian state television then played the outcome down directly, saying an agreement would not necessarily lead to an immediate opening of the strait.

The premium decayed without the constraint decaying — then partly came back the same way

This post argued a week ago that a rhetoric-driven premium is a probability estimate and decays, while a premium built on vessels actually turning around "decays only when the vessels stop turning around." The first week of August tested that and produced a genuinely awkward result twice over: roughly 10% came out of the price across two sessions with the vessels still turning around, and with a ship struck off Oman in the middle of it.

That is not a refutation of the distinction so much as a demonstration of how the two layers sit on top of each other. What the market sold was never the observed-interdiction layer — eight vessels moved through on Monday, so nothing in the physical evidence improved. It sold the forward path of that interdiction: the expected number of future days on which vessels would be turned back. An announced deal, even an unconfirmed one, shortens the expected duration of a blockade without doing anything to today's flow. Duration is a real component of a risk premium, and it is the component a headline can move.

The practical consequence is that this repricing is more reversible than the one it partly unwound. A premium anchored to observed flow requires a change in observed flow to remove it. A premium anchored to an announced framework requires only that the framework fail to appear — and the framework's own participants are publicly describing it differently. That asymmetry is the honest read, and it is not a forecast in either direction: it is a statement about what evidence would be needed to move the price back, which is a much lower bar in one direction than the other.

Thursday supplied that evidence, and it is worth noting exactly how little of it was required. No barrel changed course. No agreement collapsed — none existed to collapse. A news agency published a draft that a parliamentary commission had not yet finished reviewing, and 3.8% went back into Brent. That is the signature of a premium anchored to expectations rather than to observation: the same class of information that removed it can restore it, at the same speed, in either direction. A premium anchored to a transit count cannot behave that way, because a transit count has to be counted.

The earlier round trip in this same storyline — a 16% three-day collapse on peace hopes, then a snap-back when Iran struck three tankers — is set out in the Hormuz tanker breakdown. That episode is the reason to treat announcement-driven moves as a distinct category rather than as news about supply.

The framework now has terms — and the two sides want different ones

Monday's move traded a word. Tuesday's traded a structure, and the structure is worth reading closely, because its details are where the reopening either happens or stalls.

As reported on 5 August, what is on the table is a 60-day temporary arrangement between Iran and Oman — not a permanent settlement, and not, on Iran's account, a US–Iran agreement at all. Inbound traffic would run through a northern lane in Iranian waters; outbound traffic through a southern lane in Omani waters, coordinated with Iran. No tolls or fees would be charged during the 60 days. In parallel, the parties would work to clear naval mines from the strait's median lane within 30 days, after which that median lane would carry traffic in both directions under a permanent Oman–Iran arrangement still to be negotiated.

Two features of that design matter more than the headline. The first is that it merges two competing corridors into a sequenced hand-off, and the hand-off depends on mine clearance — a physical operation with a 30-day clock inside a 60-day agreement, in waters where a cargo ship was struck by an unidentified projectile on 4 August. The second is that the arrangement is explicitly temporary, so even full compliance restores flow for two months rather than resolving the chokepoint.

Then there is the gap between what each government has asked for, which is not a detail but the substance:

Is the strait open? There are two official answers

This is where the story becomes genuinely difficult to report, and the difficulty is itself the finding.

Rubio said on 4 August that the Strait of Hormuz "remains open and vessels are continuing to transit the waterway," while Washington worked to allow more ships to pass through safely. Associated Press reporting the same day described the strait as largely shut down as a result of Iranian attacks on ships. Both statements are defensible, because they are answers to different questions. One is about whether passage is legally and physically possible at all. The other is about how much is actually moving.

The quantities make the gap concrete, and they do not reconcile into a single tidy number:

Those two rows are not the same measurement. A vessel count includes every ship type and is sensitive to whether vessels broadcast their position; a barrels-per-day figure weights by cargo and can be sustained by a small number of large tankers. Six percent of normal traffic and thirty-five percent of normal volume can both be true at once. What neither supports is the proposition that flow has normalised.

For a factor framework this matters in a specific way. The commodities factor reads a price; the risk-sentiment factor reads the distribution around it. When the price falls 10% because the expected duration of a disruption shortened, the commodities input changes immediately while the underlying physical risk has not changed at all. That is why a currency like the Canadian dollar can watch its principal export fall by a tenth and barely move — the two factors are pulling in opposite directions, and neither has received new information about barrels.

The bypass route ends at Yanbu — and Yanbu is now a target

Every analysis of a closed Hormuz, including the one further down this page, reaches for the same consolation: there are pipelines that go around it. On 5 August that consolation acquired a specific address, and the address is under attack.

The arithmetic first. The IEA's Strait of Hormuz assessment puts total available bypass capacity at 3.5 to 5.5 mb/d. Almost all of it is one pipeline: Saudi Arabia's Petroline, the East-West crude line, which runs from Abqaiq to Yanbu on the Red Sea, has a capacity of 7 mb/d after a 2025 uprating, and carries an estimated 3–5 mb/d of spare capacity. The UAE's ADCOP to Fujairah adds up to about 700 kb/d — and Fujairah is the only one of the three that sits outside both chokepoints. Iran's Jask terminal, nominally 1 mb/d, the IEA describes as effectively non-operational and not a viable export option.

So when the world reroutes around Hormuz, the overwhelming majority of the rerouted barrels do not vanish into a pipeline and reappear on the open ocean. They come out at a port on the Red Sea, and then they still have to leave the Red Sea — either north through Suez or south through the Bab al-Mandeb strait and the Gulf of Aden.

Both of those exits are now contested. Yemen's Houthis declared a naval blockade of Saudi Arabia on 20 July. On 5 August their military spokesman, Yahya Saree, claimed missile attacks on two Saudi oil tankers — one off Yanbu in the Red Sea, one in the Gulf of Aden. The United Kingdom Maritime Trade Operations reported that a tanker sailing in the Gulf of Aden off Yemen's southern coast heard a loud explosion nearby, with all crew safe. Saudi Arabia did not confirm either incident, and Saree did not say when they took place; the Houthis have framed the campaign as a response to what they describe as a siege on Yemen, which Saudi Arabia denies.

Note carefully what this does and does not change. It does not mean the barrels stop: eight of nine claimed attacks since 20 July have not visibly halted Saudi exports, tanker attacks frequently cause no cargo loss, and a claimed strike is not a confirmed one. What it changes is the shape of the tail. Before 20 July, a closed Hormuz had a partial answer with a known capacity. After 5 August, that answer has a war-risk premium attached to its only significant exit, and the two chokepoints are no longer independent events that can be modelled separately — the same conflict drives both, and the response to one raises exposure to the other.

For a five-factor read, the consequence is narrow and specific. The commodities factor scores a price, and on 5 August the price did nothing. The risk-sentiment factor scores the distribution, and the distribution got worse while the price stood still — a second chokepoint became active on the same day the first one moved toward resolution. That combination is exactly the case in which a currency-strength framework and a price chart disagree, and it is the reason the two are kept apart. A flat settle is not a quiet day; it is two loud things cancelling.

What OPEC+ decided on 2 August

The group that met was the seven-country subset that has been setting the monthly path all year — Saudi Arabia, Russia, Iraq, Kuwait, Kazakhstan, Algeria and Oman — not the full ministerial conference. At its 5 July meeting it approved 188,000 bpd for August, the fifth consecutive monthly increase, and fixed 2 August as the next date.

On the day, Reuters reported the group was set to approve roughly 188,000 bpd for September and then hold; Bloomberg reported the seven had reached agreement in principle on the same figure. Both attributed the account to delegates. Two elements are worth separating, because they are not equally priced.

The increment itself finishes something. September restores the last of the 1.65 million bpd of additional voluntary adjustments announced in April 2023, barrels that have been returning in measured monthly slices all year. That was widely expected and is a milestone rather than a surprise.

The pause is the part that changes the framework. Holding quotas for the remainder of 2026 leaves roughly 2 million bpd of the group's older, 2022-vintage cuts in place while it works through a capacity review — a mechanism agreed last November to reassess members' maximum sustainable production, covering 19 of the group's 22 members and feeding the baselines from which 2027 quotas will be set. Iraq and others want individual quotas that reflect expanded capacity. Those are difficult talks, and until they conclude, the alliance has moved from executing a schedule to exercising discretion.

The month the war premium became a shipping story

The sequence matters, because each leg was driven by something different, and a five-factor read scores them differently.

Into 27 July, Brent fell about 16% across three sessions — its steepest three-day decline since 2020 — as a pause in the US–Iran fighting appeared to hold and crude slid back below $90. That was a forecast changing: the market removed a probability-weighted disruption it had been carrying.

Then it came back. On 28 July Iran's Revolutionary Guard fired ballistic missiles at US forces in the region; US Central Command said all were intercepted with no casualties or damage. Iran-aligned militias in Iraq launched drones at oil facilities in Saudi Arabia's Eastern Province for a second consecutive day. On 29 July, after President Trump said Iran was "going to get a beating" and that the US would be "hitting them hard," Brent settled 7.9% higher at $90.74 and WTI rose 6.6% to $84.46.

The final two sessions are where the character of the move changed. Reuters reported that Iran's Revolutionary Guards had halted two tankers and forced four others to change course in the Strait of Hormuz. Brent settled Friday 31 July up 1.2% at $90.12 and WTI up 1.3% at $84.67, capping monthly gains of 24% and 21% respectively — the largest since March. Ole Hvalbye, a market analyst at SEB Research, put the shift plainly: "The market has stopped trading the war and started trading the shipping data."

That was the distinction this post was built on, and the first week of August is what tested it. The daily benchmark series are on FRED for Brent and WTI.

The scenario that landed, factor by factor

The preview mapped three branches. The first one landed — and the reason it produced so little immediate currency impulse is instructive.

The base case is the branch with the least immediate price content and the most structural content. Nothing about a pre-announced 188,000 bpd repriced anything on Sunday. But the alliance has now spent its visible supply, which means that for the rest of 2026 the oil price is set by two things a producer group does not control — the physical flow through Hormuz and the level of inventories — plus a discretionary reserve nobody can schedule. Uncertainty of that kind does not show up as a one-day move. It shows up as a wider distribution around every subsequent headline, and in a factor framework that reaches currencies through the risk-sentiment channel before it reaches them through commodities.

Why the barrels were never the point

Set the quota against the risk it is being asked to offset. The IEA reports that 20 million barrels a day of crude and oil products transited the Strait of Hormuz in 2025 — around 25% of the world's seaborne oil trade, of which nearly 15 mb/d was crude, some 34% of global crude trade. Just over 112 bcm of LNG also moved through it, almost 20% of global LNG trade.

Against that, the alternative routes are thin — and, as set out above, mostly exposed. The IEA puts total available capacity on bypass pipelines at 3.5 to 5.5 mb/d: Saudi Arabia's Petroline to Yanbu on the Red Sea with an estimated 3–5 mb/d of spare capacity, the UAE's ADCOP to Fujairah with up to 700 kb/d of additional volumes, and Iran's Jask terminal effectively non-operational despite its nominal 1 mb/d. Only the ADCOP volumes clear both chokepoints.

A 188,000 bpd quota step is under 1% of what passes through the strait each day. It cannot insure against the tail; it was never designed to. What it does do is exhaust the group's pre-committed supply — and it did so in the same week that vessels were physically turned back in the strait it cannot bypass.

The importers got their confirmation — in writing

The cleanest fundamental effects of the oil price are not in Calgary. They are in Tokyo and Frankfurt, and Tokyo made the mechanism explicit the day before the meeting — while Brent was still near $90, which is what makes the subsequent fall a live test of its forecast rather than a footnote to it.

The Bank of Japan held its policy rate at 1% on 31 July in an 8–1 vote, with board member Hajime Takata proposing a hike to 1.25%. The interesting part was the quarterly Outlook for Economic Activity and Prices, which said core inflation was likely to accelerate to a level clearly above 2% from the second half of fiscal 2026. Among the reasons it gave: wage increases being passed into selling prices, the recent depreciation of the yen, and the rise in crude oil prices, which it expects to push up energy and goods prices. Its projection then has inflation easing back toward 2% as crude declines — an assumption when it was written, and one the first week of August has begun to supply. Brent at $79.36 is a materially different import bill for a near-total crude importer than Brent at $90.12, and if it holds it removes one of the three reasons the BoJ gave for inflation running clearly above 2%. Whether it holds is a question about the strait, not about Japan — and Thursday's close at $82.49 is a reminder of how provisional the answer is. Roughly half of the relief the first three sessions of August delivered to Japan's import bill was withdrawn by a draft bill, which is a fair summary of how much of this inflation path currently rests on a shipping lane.

That is a central bank writing an oil price into a policy forecast. It is the same terms-of-trade deterioration this site traced in the rate-gap breakdown, now layered on a currency already at four-decade lows. The eurozone faces the same import bill arriving through headline HICP, where the July flash estimate rose to 2.9% — covered in the flash CPI breakdown.

Then there is the grouping error. "Commodity currencies" is a useful bucket, but it is not an energy bucket. Canada is a large net crude exporter. Australia is a substantial energy exporter through LNG and coal, though far more levered to Chinese demand and global risk appetite than to crude. New Zealand is a net importer of crude and refined fuel — for the kiwi, $90 Brent was a cost rather than a windfall, and the fall to $79.36 is relief rather than a hit. The underlying mechanics are set out in the terms-of-trade explainer, with live reads on the AUD and NZD pages.

The loonie's fortnight, in official data

The preview asked whether USD/CAD would finally respond to oil. Across three weeks it did not — and then it responded to something else entirely on 7 August. The record is unusually clean because the fortnight contained every kind of shock at once, in both directions, and then a control experiment at the end of it.

Those are Bank of Canada daily rates. Across five sessions the loonie gained about 0.6% against the dollar — a range that would be unremarkable in a quiet week, produced here by a week that was anything but.

The domestic data did not move it either. Statistics Canada reported on 31 July that real GDP by industry rose 0.3% in May, above the 0.2% consensus, with 13 of 20 sectors contributing, goods-producing industries up 0.6% and mining, quarrying and oil and gas extraction up 1.0%. The advance estimate put June at +0.2% and the second quarter at roughly 0.8%. A growth beat and an oil rally in the same week, and USD/CAD ended it near where a quiet week would have left it.

The following three sessions closed the loop from the other side, and the official series is unusually easy to read because one of the days does not exist. No Bank of Canada rate was published for Monday 3 August — it was the Civic Holiday. The next official observation, Tuesday 4 August, was 1.4068, so across two sessions in which Brent fell 5.1% and then 5.3% the Canadian dollar weakened by 0.28%. Then Wednesday 5 August printed 1.4026 — back through the 31 July level.

Thursday 6 August then printed 1.4018 — a 0.06% gain for the loonie on a session in which Brent rose 3.8%.

That is the number to sit with. Between Friday 31 July's 1.4029 and Thursday 6 August's 1.4018, the Canadian dollar moved 0.08% against the US dollar. Effectively nothing. In between, Brent gave back roughly a tenth of its value and then recovered nearly 4% of it, a 24% monthly gain in Canada's principal export was substantially unwound, a second maritime chokepoint became active, and a Canadian GDP beat landed at the front of the window. Three weeks of the most violent tape crude has produced since 2020 have now delivered a cumulative currency move smaller than a normal morning's drift.

Read strictly, this is not a story about oil failing to matter. It is a story about a variable being crowded out. The loonie declined to respond to a 16% three-day crash, a 7.9% single-session rally, a 24% month, a GDP beat, a two-session 10% slide, a round trip back to flat and a 3.8% rebound — in both directions, across three consecutive weeks. When a relationship fails to fire in both directions on large moves, the honest conclusion is not that the sign is wrong but that the channel is not the binding one. What 7 August added was the other half of the proof, because a channel that is not binding is only an interesting claim if some other channel demonstrably is. The full record sits in the same daily series.

The next scheduled test of which factor is actually binding arrived on Friday 7 August with the July Labour Force Survey — and it answered the question in a single session. Canada added 75,100 jobs and the unemployment rate fell to a two-year low of 6.4%, set out in full in the Canada jobs breakdown. USD/CAD's official Bank of Canada rate closed at 1.3943, from 1.4018 the day before: a 0.53% gain for the Canadian dollar, the largest single-day move of the entire three-week episode. Brent that day moved 0.45%, in the other direction, and Iran declared its Hormuz framework finalised.

The test was designed to be a low bar and it cleared it by a distance. A labour-market print moved the loonie roughly six times further in one day than a complete round trip in crude — a 24% monthly gain, a two-session 10% slide and a 3.8% rebound — had moved it across the preceding week. State that as a finding rather than a surprise: for CAD in August 2026 the rate channel is the binding factor and the commodities channel is not, and the evidence is now symmetric. Oil moved violently and the currency did nothing; the labour market moved moderately and the currency moved. That is what a crowded channel and an uncrowded one look like side by side, and it is the distinction a factor framework exists to make rather than to assume.

There is a mechanical reason the symmetry is this clean, and it is worth stating because it is easy to mistake for a broken relationship. A cheaper barrel weakens CAD directly through the commodities factor and through Canada's terms of trade. But the same cheaper barrel is risk-on for a pro-cyclical currency, which is CAD-positive, and it softens the US inflation impulse, which weakens the counter-currency in the pair. Three channels, two of them pointing the other way. When one catalyst moves all three at once, the net is close to zero — not because the oil–CAD link has failed, but because it is crowded.

Monday 10 August then delivered the confirming observation, and it is the tidiest one in the whole record. Brent settled about 5% higher at $87.72 — the largest single-session gain since 29 July — and USD/CAD's official Bank of Canada rate printed 1.3942 against 1.3943 on Friday. One basis point of currency on five hundred basis points of oil. Read the two adjacent sessions together and the experiment is about as controlled as macro ever gets: a domestic labour-market print moved the loonie 0.53%, and the next session's 5% move in Canada's principal export moved it 0.01%. Fifty times the response from a fraction of the shock, in the channel this post has been arguing is binding.

The conclusion is the same one the preview reached, now with a fuller data set and a clean counterfactual behind it: the commodities factor is not the marginal driver of CAD at present. The rate gap is, as covered in the Canada GDP preview and in the oil–CAD relationship in full. When a currency ignores a 24% rise in its principal export and then a two-session 10% fall, and then moves 0.53% on a labour-market print, that is information about which factor is binding, not noise.

What the pause changes from here

Three things, in order of how much they will actually move a factor score.

First, whether transit volumes actually recover — and, after 7 August, whether the transit can be insured at all. This was the top item before Monday, and it has now returned several numbers, all of them low and the most recent ones falling: eight vessels on 3 August, a range of 8 to 15 across 4–6 August, then 15 on Friday 7 August, 11 on Saturday and 6 on Sunday, against roughly 130 a day pre-crisis, with crude and product flow near 7 mb/d against about 20 mb/d in 2025. Roughly 10% came out of the price on an announced reopening and about 3.8% went back in when the reopening's draft terms were published, and the announcement has so far produced documents rather than tankers. The Lloyd's clause changes the shape of this item rather than its position on the list: the question is no longer only whether a deal is signed but whether the signed version contains a fee, because a fee-bearing deal cannot be used by insured tonnage regardless of who signs it. Watch for the fee to be dropped, for the collecting entity to be de-designated, or for underwriters to revise the clause. Absent one of those three, a signature is not a shipping lane. That does not make the repricing wrong — a 60-day arrangement with a 30-day mine-clearance clock is a real path to higher flow, and if it is signed and executed the market will have been early rather than mistaken. But the verification is a transit count, it is observable within days, and it has not yet moved. If the framework stalls, the premium has a clear route back into the price, because nothing physical was ever removed. Watch flow data, not statements — the statements are precisely what is in dispute, to the point where two officials described the same strait as open and as largely shut on the same day.

Second, whether the premium starts feeding inflation forecasts elsewhere. The BoJ has already done it, and the two-session slide cuts both ways here: the BoJ's own projection has inflation easing back toward 2% as crude declines, so a sustained fall would validate an assumption it has already banked, while a reversal would leave it underwriting a forecast the oil market has stopped supporting. If Brent instead settles back near $90 into the autumn, the same arithmetic reaches eurozone headline HICP and, more slowly, US energy CPI — into a Federal Reserve that has just withdrawn its own forward guidance, covered in the 9–3 hold breakdown. With no guidance to filter incoming data, an energy-led inflation path transmits to the dollar's rate factor faster than it used to.

Third, whether the 2027 baseline talks leak. The capacity review is the mechanism that decides how much supply exists on paper from January 2027, and reporting on who is arguing for what will move the forward curve well before any decision is taken. That is a genuinely new source of headline risk, and it replaces the monthly quota as the thing to watch. The earlier phase of this unwind, when the story was a glut rather than a shortage, is in the August output-hike breakdown.

The first week of August added a fourth item that sits underneath all three: the alliance now has no scheduled supply to answer a reversal with. That is the part of the pause that only shows its shape when the price moves. Had the framework arrived in June, OPEC+ still had four monthly slices in hand and a visible schedule the market could read. From October it has neither. So if the announced reopening does not produce tankers, the group's response is discretionary, unscheduled and negotiated during a conflict — and a market that has just taken 10% out of the price on an announcement would be re-pricing that risk from a lower base, with a thinner buffer behind it.

And 5 August added a fifth, which may outlast the other four: the bypass now needs its own risk premium. Every scenario in which Hormuz stays shut runs through Yanbu, and Yanbu sits inside a declared blockade zone whose operators have said they will escalate precisely because barrels are being diverted there. That is not a reason to expect any particular price. It is a reason to stop treating the two chokepoints as separate lines in a resilience table, because the mitigation for one is the exposure to the other. If the Hormuz framework is signed and executed, this matters much less. If it stalls, the fallback is thinner than 3.5–5.5 mb/d makes it look. On 9 August it stopped being a projection: the Jizan refinery, 400,000 bpd on the Red Sea coast, was hit by a drone, and the escalation moved from vessels to fixed plant.

The weekend of 8–10 August then added a sixth, which sits underneath all five: the non-market buffer is nearly spent. The Strategic Petroleum Reserve has supplied roughly 108 million barrels since April, about 900,000 barrels a day, and at 304.8 million barrels it stands at a level last seen in February 1983. That release has been absorbing part of the Hormuz shortfall without appearing in any production figure, which is one reason the price has spent the summer trading documents rather than scarcity. Watch the weekly EIA number — the next release is 12 August — because a drawdown that slows or stops removes a supply source the market has been receiving quietly and for free. Combined with the OPEC+ pause, both of the system's discretionary buffers are close to exhausted in the same month. That does not imply a direction for the price. It implies a wider distribution around whatever direction arrives, and that reaches currencies through the risk-sentiment factor before it reaches them through commodities.

None of this is a forecast. It is a decomposition: what the group actually decided, how large that decision is relative to the risk it sits inside, and which of the five fundamental factors carries the result into each currency. On Sunday 2 August the headline was a number of barrels, and the number of barrels was not the story — the exhausted schedule behind it was. On Monday the headline was a deal, and the deal was not the story either. On Tuesday the headline was the deal's terms. On Wednesday the coordinates were agreed and the price closed seven cents higher, because in the same session a missile was aimed at the alternative. On Thursday the terms were published in draft, they barred the ships of the country negotiating for their removal, and 3.8% went back into the price. On Friday the framework was declared finalised and Brent fell 0.45%, because the binding constraint had migrated out of the negotiation and into an insurance clause and a sanctions list. On Saturday the constraint moved again — into sanctions policy, frozen assets and reparations — and on Monday Brent settled 5% higher at $87.72. Eight sessions, eight different documents, and not one additional barrel.

That progression is the finding, and it is more useful than any level in it. Each time the obstacle has moved, it has moved to a venue further from the water: from vessels turned back, to a negotiating room, to an underwriter's clause and a sanctions list, and now to a reparations claim that both sides are making against each other. A premium anchored to the first of those is removed by counting ships. A premium anchored to the last cannot be removed by anything observable on a weekly schedule — which is why the price keeps travelling 5% and 10% in either direction on paperwork while the strait passes single-digit numbers of vessels.

Three numbers close the fortnight, and they point in different directions about what deserves attention. The first is the one that will settle the oil question and has now moved the wrong way: 15 ships on Friday, 11 on Saturday, 6 on Sunday, against roughly a hundred and thirty a day before the conflict began on 28 February. What the strait is actually passing — and what leaves Yanbu and Jizan — is still the only evidence that cannot be renegotiated, voided by an underwriter or attached to a sanctions demand. The second is 304.8 million barrels, a level the Strategic Petroleum Reserve last saw in February 1983, which is the answer to what has been quietly absorbing the shortfall and how much of that answer is left. The third is 1.3942, which is where the Canadian dollar sat on a day its principal export rose 5% — one basis point from where a jobs report had left it on Friday. A fortnight that produced this much oil news and moved this currency only through a labour-market print is a reasonably direct answer to the question of which factor is carrying the result.

For more on how currency strength is built from fundamentals rather than price, see the about page, and for the loonie specifically, the CAD currency page.

Educational macro context only — not investment advice.