Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- Build a one-page decision sheet this week pricing one attempt against one completed voyage; with six attempts and zero completions on the record, treat any single transit as a binary bet, not a rate.
- Recompute exposure in tonne-miles rather than barrels: at 20% of volume lost and a 25% longer haul, demand for ships is flat, and every percentage point of detour beyond that adds shipping demand instead of removing it.
- Check days of crude cover at every receiving site before month end; anything under 45 days moves to a weekly review, because 20 extra voyage days consumes two thirds of a 30-day cushion.
- Ring-fence the fuel pass-through now: model one bunker adjustment in the next index cycle and put a ceiling into customer quotes rather than absorbing it after the fact.
- Require two independent sources before any routing decision through the strait; the only casualty figure in this story comes from one belligerent, so a model built on it prices a claim and calls it a risk.
Iran's Revolutionary Guard said on Friday 31 July 2026 that it struck two oil tankers attempting to transit the Strait of Hormuz under US military aerial escort, describing the vessels as non-compliant, and said four additional tankers turned back. US Central Command said on 30 July it had redirected 24 commercial vessels since resuming a blockade of Iranian ports earlier in the month. Iran also said it launched drone attacks on the Ahmad al-Jaber Air Base in Kuwait, where the defence ministry reported intercepting hostile drones and material damage from falling debris but no casualties. The strait, which normally carries about 20% of global oil shipments, has been effectively closed since the war began on 28 February.
The Analysis
Six ships tried. Two were hit. Four turned around. Nothing arrived. That is the only set of numbers in the 31 July statement that can be used to price risk, and every headline I read led with a different one: twenty percent of global oil shipments.
Twenty percent is a share of a flow. It is not a probability, not a rate, and not the denominator an owner needs tonight when he decides whether to sail. Turning a share into a probability is the most common arithmetic accident in this kind of story.
So this is one exercise: put the three numbers back on their own rulers.
Three numbers sit in this story and no two of them are on the same scale. Twenty percent is a share of a flow. Twenty-four is a count of instructions issued since a blockade resumed, with no total of how many ships presented themselves. Two is a hit count. Adding them, comparing them, or converting any of them into a probability is how bad risk pricing gets made.
Take the two first. Two tankers struck while attempting to transit under aerial escort, four more turned back. Six attempts, two hits, four refusals, zero completions. The hit rate on the attempt is one in three. It is not two in twenty-four, and it is not two in the thousands of transits the strait used to handle in a month.
The denominator is the argument. An owner testing the closure is not sampling from the population of normal transits. He is sampling from the population of ships that decided to try, and that population is six. A small sample is not a statistical inconvenience here, it is the actual decision environment. Nobody gets a thousand observations to average. You get one attempt and a binary outcome.
Inside that denominator sits another. Four ships turned back. They burned fuel, paid crew, paid for escort coordination, and delivered nothing. Their cost per delivered barrel is not high, it is undefined, because the denominator is zero. That cost does not disappear. It gets pushed onto the next fixture that completes, which is one reason the price of barrels that move is a poor guide to the price of barrels that do not.
Twenty percent of global oil shipments is a volume share, and volume is the least interesting property of a chokepoint. What matters is substitutability: whether the barrels that cannot pass can be replaced by barrels from somewhere else. If they can, the market reprices freight and the crude travels further. If they cannot, the market reprices everything else.
The arithmetic that gets skipped is tonne-miles. Freight demand is barrels multiplied by distance. Lose a fifth of the barrels and route the surviving four fifths 25% further — my assumption, not a figure from the report — and tonne-mile demand is exactly unchanged, because 0.8 times 1.25 is 1. Route them 40% further and it rises 12%. A blockade that removes cargo can increase the demand for ships. That is not a paradox, it is multiplication.
Who carries it. A refiner with tankage and a term contract can wait out a month. A refiner running hand to mouth cannot, and pays whatever the spot voyage costs that week. On the ship side, an owner whose vessel is already inside the Gulf is not choosing a rate, he is choosing whether to sail at all, and that answer comes from the insurer, not the charterer.
Container shippers are one step removed and should not feel relieved. The link is fuel. Bunker is a cost that passes through on a schedule, and a crude event that moves the fuel curve reaches a container invoice through an index, on the index's timetable rather than yours. By the time it arrives, the cause is three headlines old.
The first invoice to move is the next fixture, not the cargo at sea. War-risk cover is quoted per voyage, so an owner re-enters the market every time he fixes; cargo already loaded keeps the cover it was booked under. The price signal reaches the spot market in days and the contract market in months, and the gap between them is where money is made and lost.
What the coverage missed is the asymmetry between two instruments. A strike is an event with a date. A closure is a condition with no date. Insurance prices events, not conditions, because a condition has no end point to discount against. Two ships hit is a claim. Twenty-four ships redirected is an operating environment. The first has a market. The second has a queue.
That distinction changes what you can buy. You cannot insure a queue. You can only shorten the time your cargo spends inside the affected water, and the lever for that is the date on your booking, not the rate you pay to be there.
Run the buffer arithmetic with my assumptions flagged. A refinery holding 30 days of crude cover, against a rerouted voyage that adds 20 days, has two thirds of its cushion eaten by the first cargo that goes the long way. It does not run dry. It runs thin, and thin is where panic buying starts, and panic buying is what turns a shipping event into a price event. The number to watch is not the global oil share. It is days of cover at the point of consumption, and nobody prints that in a headline.
If the strait had stayed open, that twenty percent would have moved at short distance, tonne-miles would have stayed flat, and the freight market would have shrugged. The closure did not destroy the oil. It stretched the map. That is why the effect shows up in ship demand before it shows up in shortage.
The trap is reading twenty percent as your own exposure. If you import containers, your direct exposure to this strait is close to zero; your indirect exposure is the fuel line and the insurance line, both of which move on schedules you do not control. The second trap is treating a claim as a finding. The hit count comes from one side of a war. There is no independent number in the story, and a risk model built on a belligerent's press release is a model of a press release.
Six attempts, two hits, four refusals. The strangest part of this market is that the safest ship in the Gulf tonight is the one that turned around with an empty tank and an intact hull. It earned nothing and lost less than the two that tried.
↑ Back to the key points