The Probability Desk · Wednesday, 9 September 2026 · Universal Asset Owners
Brent crossed $100 this morning for the first time since July. Five named houses have published a forecast for the same month of 2027 within the last three weeks, and on a common basis they are $22.40 a barrel apart — the printed numbers themselves span $30. That spread is not a disagreement about oil. It is a disagreement about whether the world's most important chokepoint reopens free, reopens metered, or does not reopen at all — and only one of those three outcomes is in the official number.
Executive summary
Yesterday U.S. Central Command destroyed five Iranian crude tankers in the Gulf of Oman and near Kharg Island, the second such action in four days, after the Islamic Revolutionary Guard Corps tried twice in two days to hit a U.S. Navy warship with ballistic missiles. Iran fired ballistic missiles at an air base in Jordan; Jordan's military said it intercepted eighteen of them. The IRGC told tanker crews at Kuwaiti and Bahraini piers to abandon their vessels. Brent, which settled around $98, moved above $100 as reports of explosions near Kharg circulated.
None of that is the story for capital that thinks in decades. The Strait of Hormuz has been impaired since 28 February. Traffic is running at roughly 8 million barrels a day against a fiscal-2025 baseline of 19.87 million. Qatar has shipped eighteen LNG cargoes in six months of war against 509 in the same period a year earlier. The market has had six months to price a constrained Strait, and it has.
The story is what the reopening looks like. On 26 August, Iran's deputy foreign minister announced that Iran and Oman had agreed a temporary seven-mile maritime corridor through Iranian territorial waters, transited only with Iranian permission. An IRGC spokesperson said in the same breath that agreements had been reached on "the share of Iran and Oman in its revenues." A senior Iranian official has floated a transit fee of 5–7% of cargo value; flat-fee alternatives reported at around $1 a barrel, equivalently $1–2 million per vessel per voyage, are also in circulation. Oman has counter-proposed a voluntary contribution modelled on the Strait of Malacca. The United States rejects any charge on what it calls a once-free international waterway.
A quarter of the world's seaborne oil trade, and a fifth of its LNG, have moved through that water for free for as long as there has been a market in either. The live question is no longer whether the Strait closes. It is whether it comes back metered.
The Desk's weights for the March 2027 Brent monthly average: The Reopening (under $75) 25% · The Impaired Grind ($75 to under $95) 40%, the base case · The Toll Regime ($95 to under $125) 30% · The Break ($125 or higher) 5%. P(≥ $95) = 35%.
The U.S. Energy Information Administration published its Short-Term Energy Outlook this morning. It puts Brent at exactly $83 in March 2027. That number sits inside our base case. We do not think it is wrong. We think it is a line where the question is a distribution — and that the 35% of probability mass above $95 is the part of this trade that no point estimate can carry.
The Desk view in one paragraph
The consensus forecast for oil in early 2027 is not a forecast about oil. It is a reopening schedule wearing a price forecast's clothes. The EIA's own numbers make this explicit: its balance sheet swings from a 1.71 mb/d inventory draw in Q4 2026 to a 3.01 mb/d build in Q1 2027 — a 4.7 mb/d reversal inside a single quarter — and that swing, not any view about demand, is what produces the glide to $83 and then to $62 by December 2027. The swing requires shut-in barrels to come back. Shut-ins were 6.7 mb/d in August, up from 5.0 mb/d in July. Nothing that happened this week resembles a Strait on a Q1 restoration schedule. We are therefore above the official path in the upper half of the distribution and, unusually, close to it in the middle: our median is $81 against the EIA's $83. The difference between the Desk and the consensus is not the central estimate. It is that we assign 35% to an outcome at or above $95, and the consensus assigns a point.
The situation as of today
The war began on 28 February 2026 and has already round-tripped once. A U.S.–Iran memorandum of understanding signed 17 June produced a sixty-day arrangement in which Iran agreed to keep the Strait open free of charge; Brent fell to $69 on 2 July. The ceasefire collapsed on 8 July, the United States reimposed the naval blockade in mid-July, and Brent spiked to $105 on 23 July. This is the second inflation of the same premium, not the first — a fact that matters enormously for how any decay model is specified, and one that a single-start-date half-life estimate gets wrong by construction.
What is measurably true today:
- Transit is impaired, not stopped. Vortexa put Hormuz liquids at roughly 8 mb/d on a 28-day moving average to 31 August. TankerTrackers put crude specifically at 5.94 mb/d on a seven-day average. Kpler counted 6, 11 and 5 vessel transits on three consecutive days at the start of September against a ten-day average of 13. HSBC's note of 8 September says flows have "settled around 30% of pre-conflict levels." The U.S. Department of Energy told Axios that about 12 mb/d moved on Wednesday 2 September, and the President posted that "18 million barrels per day are moving." Drewry's Eirik Hooper explains the gap: an operational count plausibly includes naval auxiliaries, tugs, offshore support and dhows, where the commercial trackers filter on cargo and size. We use the commercial trackers, and we flag that by late August the majority of crossings were classified dark or unknown by route.
- The bypass is itself under attack. The IEA's factsheet puts alternative crude export capacity around Hormuz at 3.5–5.5 mb/d, principally the Saudi East–West line to Yanbu and the UAE's Habshan–Fujairah pipeline. Saudi Arabia pushed East–West to a record 7 mb/d in Q1. But the EIA's outlook this morning reports that August loadings out of Yanbu fell by about half from July because of attacks on Bab el-Mandeb, with cargoes shifted to Suez and to ship-to-ship transfers outside the Gulf. Treat the bypass number as paper capacity, not operating capacity.
- LNG is a stoppage, not a constraint. Qatar loaded 18 cargoes in the first six months of the war against 509 a year earlier, a decline of roughly 96%, and about $24 billion of lost gas sales. QatarEnergy's force majeure runs into early November. Non-Gulf LNG output rose about 18% over March–June and offset roughly three-quarters of the Gulf shortfall; global LNG production fell about 4%.
- The freight market has gone somewhere the crude market has not. The Baltic Exchange's tanker report of 4 September assessed TD3C — 270,000 tonnes Middle East Gulf to China — at Worldscale 677.22, a round-trip time-charter equivalent just under $704,000 a day, against WS126.5 and $112,394 a day in the same report in January. That is Worldscale up 5.4× and TCE up 6.3×. Lloyd's List has separately cautioned that TD3C is now assessed on a route many owners will not physically sail; the economically live index is the Gulf-of-Oman loading, TD34, at WS272.5 and about $261,800 a day. We report both and use neither as an achievable earnings number.
- Credit has barely moved, and the exception is not the one people name. Investment-grade GCC five-year CDS rose only 7–23 basis points to their March peaks and have since retreated. Year-to-date to 24 August: Saudi Arabia −9 basis points, UAE +6, Qatar +6, Kuwait +4, Oman +7. The outlier is Bahrain, +101 basis points on the year at 284. Saudi credit is tighter than it started the year. Any account of this war that has Gulf sovereign risk repricing has not looked at the prints.
The forecast question
Where will the Brent crude monthly average stand in March 2027?
Resolution source: the published value of FRED series MCOILBRENTEU (Crude Oil Prices: Brent — Europe, monthly, sourced from the U.S. Energy Information Administration) for the period 2027-03-01. Sole arbiter. No judgement required. Later revisions do not change the grade.
Buckets, mutually exclusive and collectively exhaustive:
| Bucket | Range (monthly average, $/bbl) | Weight | |
|---|---|---|---|
| A | The Reopening | under $75.00 | 25% |
| B | The Impaired Grind | $75.00 to under $95.00 | 40% |
| C | The Toll Regime | $95.00 to under $125.00 | 30% |
| D | The Break | $125.00 or higher | 5% |
Starting level: $91.08, the August 2026 monthly average — the last complete published observation and the same unit as the resolution variable. Spot has since traded above $100; that gap is carried in the simulation, not in the base rate, and the distinction is stated wherever it matters.
Horizon: seven months. P(≥ $95) = 35%.
Prior and analogues
The reference class is physical supply shocks that removed barrels or threatened a chokepoint, measured on the same variable and the same seven-month horizon. We computed the outcomes rather than asserting them; base_rate.py ships with this report and runs offline from its own data slice.
| Episode start | Event | Brent then | Brent +7 months | Change |
|---|---|---|---|---|
| Aug 1990 | Iraq invades Kuwait | $27.17 | $19.08 | −29.8% |
| Dec 2002 | Venezuela PDVSA strike | $28.33 | $28.35 | +0.1% |
| Mar 2003 | Iraq war begins | $30.61 | $29.61 | −3.3% |
| Feb 2011 | Libya civil war | $103.72 | $112.83 | +8.8% |
| Jul 2012 | EU embargo on Iranian crude | $102.62 | $116.05 | +13.1% |
| Sep 2019 | Abqaiq/Khurais attack | $62.83 | $18.38 | −70.7% — see below |
| Feb 2022 | Russia invades Ukraine | $97.13 | $89.76 | −7.6% |
| Jun 2025 | Israel–Iran 12-day war | $71.44 | $66.60 | −6.8% |
One of these eight windows is contaminated, and we caught it because the class mean looked implausible. The seven months after Abqaiq end in April 2020. That −70.7% is the COVID demand collapse, not the decay of a drone-strike premium. Including it moves the class mean from −4.6% to −17.7% and bucket A from 14.3% to 25.0% — a single bad row moving the class by more than ten points. It is excluded from the blend and reported here so the exclusion is auditable rather than invisible.
Three deeper points from the analogue set:
The Tanker War is the closest analogue and it says the shooting does not set the price. Between 1984 and 1988, 239 petroleum tankers were attacked in the Gulf, of which 55 were sunk or written off; tankers were 61% of all ships attacked. The Strait never closed. Citing Blair and Lieberthal in Foreign Affairs, the Strauss Center records that even at its most intense the campaign "failed to disrupt more than two percent of ships passing through the Persian Gulf." Across those same four years the real oil price fell, culminating in the 1986 collapse from $23.29 to $9.85 on the OPEC basket — a collapse driven by Saudi abandonment of the swing-producer role, not by the war. Physical attrition of tankers is economically trivial. Removed barrels are not. That is the distinction on which this entire edition turns.
The premium decay literature is short and consistent. The ECB's Economic Bulletin 8/2023, in a box by Ferrari Minesso, Lappe and Rößler built on the Caldara–Iacoviello geopolitical risk index, finds that "oil price pressures arising from adverse geopolitical shocks are generally short-lived, with elasticities becoming insignificant after one quarter for most countries." Russia is the single exception at about +2% beyond a quarter, which the authors attribute to the import embargo — a policy channel, not a risk premium. The same box records the Ukraine invasion as +30% in two weeks and back to pre-invasion levels in roughly eight weeks. The Dallas Fed's Kilian, Plante and Richter find that a 20-percentage-point rise in the probability of a 5% oil production shortfall reduces output by only 0.12%.
But 2026 is the one case in the modern set where the barrels actually left. The Tanker War, the Red Sea campaign and the 2025 twelve-day war all left the oil flowing; what repriced was freight, insurance and voyage days. Here transit is running at roughly 40% of baseline and shut-ins were 6.7 mb/d in August. The two episodes where a premium genuinely persisted — 1973 and 1979 — did so because prices were administratively set, there was no non-OPEC swing supply, demand could not respond within the horizon, and OPEC was the price-setter. None of those four conditions holds now. So the historical prior points down, and the physical situation points up, and the honest answer is that both are true and the distribution is wide.
Base-rate leg: A 15.8 · B 36.8 · C 40.5 · D 6.9. Note that history is the most bullish of our three legs, not the least. A six-month run of +28.5% into the 89th percentile of its own ten-year range has historically stayed elevated.
The evidence-update table
No probability in this report moves without the evidence that moved it. Directions are relative to the base-rate leg above.
| # | Evidence | Source & date | Direction | Strength | Effect on the model |
|---|---|---|---|---|---|
| 1 | Five IRGC tankers destroyed 8 Sep; second action in four days; IRGC threatens crews at Kuwaiti and Bahraini piers | CENTCOM release, 8 Sep 2026; Stars and Stripes, 9 Sep | ↑ C/D | Medium | Raises the escalation hazard; does not by itself remove barrels |
| 2 | Iran–Oman corridor agreed 26 Aug, seven miles wide, Iranian permission required, revenue-sharing explicitly discussed | Al Jazeera, 26 Aug 2026, quoting Dep. FM Gharibabadi and an IRGC spokesperson | ↑ C, ↓ D | High | Creates the toll channel: a reopening that leaves a permanent per-barrel wedge |
| 3 | Transit ~8 mb/d vs 19.87 baseline; ~30% of pre-conflict per HSBC | Vortexa via Axios, 4 Sep; IEA factsheet, Feb 2026; HSBC note, 8 Sep | ↑ B/C | High | Sets the model's initial state; the shortfall is the price driver |
| 4 | EIA balance flips from a 1.71 mb/d draw in Q4 26 to a 3.01 mb/d build in Q1 27 | EIA STEO series T3_STCHANGE_WORLD, released 9 Sep 2026 |
↓ A | High | The official case is a reopening assumption; identifies what must be true for A |
| 5 | Shut-ins 6.7 mb/d in August, up from 5.0 in July; forecast 5.7 mb/d in Q4 | EIA STEO, 9 Sep 2026 | ↑ B/C | High | Shut-ins are rising into the quarter in which they are forecast to fall |
| 6 | Yanbu loadings down ~50% m/m in August on Bab el-Mandeb attacks | Vortexa, cited in EIA STEO, 9 Sep 2026 | ↑ C | Medium | The bypass is degraded; paper capacity ≠ operating capacity |
| 7 | Kharg loadings 1.98 mb/d in Feb → 135,000 b/d in August | Kpler, via The National, 31 Aug 2026 | ↓ D | Medium | Iranian export capacity is already almost entirely out; less left to lose |
| 8 | IEA released 400 mb in March, the largest collective action in its history | IEA, 15 Mar 2026 | ↓ C/D | Medium | Buffer already substantially spent — a used shock absorber |
| 9 | IEA sees 2027 supply exceeding demand by more than 5 mb/d | IEA Oil Market Report, Aug 2026 | ↓ A/B floor | High | Anchors the fundamental level below the current price |
| 10 | GCC investment-grade CDS flat to tighter; Saudi −9 bp YTD | AGBI, 28 Aug 2026, citing Arqaam and UBP | ↓ D | Medium | Credit is not pricing a systemic Gulf event |
| 11 | The June MoU already produced a full round-trip to $69 and then failed | EIA STEO, Aug 2026 | ↑ variance | High | A settlement is achievable and fragile: widens both tails, favours neither |
| 12 | VLCC TD3C TCE ~$704,000/day vs $112,394 in January | Baltic Exchange, 4 Sep 2026 | neutral on flat price | Medium | Confirms the Red Sea lesson: chokepoint stress reprices freight first |
Net: evidence rows 2, 4, 6 and 11 are the ones that actually move this model. Rows 4 and 6 push against bucket A; row 2 opens bucket C as a structural rather than merely a conflict outcome; row 11 widens everything.
The scenarios
A — The Reopening · under $75 · 25%
The Oman corridor is formalised without a mandatory toll, or with a token voluntary contribution on the Malacca model. Iran accepts a package restoring elements of the lapsed June memorandum; the naval blockade is lifted; shut-in barrels return over three to four months into a market the IEA already expects to be more than 5 mb/d oversupplied in 2027. Prices do not merely retrace, they undershoot, because the 2027 surplus was always going to arrive and the war deferred it rather than cancelled it. Citi is explicitly here: its 3 September note keeps 2027 at $65 on the reasoning that a reopened market is in a larger surplus than the pre-conflict one.
What puts it in play: a signed successor to the June MoU with sanctions relief attached. Tripwire: Hormuz transit above 14 mb/d on a 28-day Vortexa average for four consecutive weeks. What falsifies it: shut-ins failing to fall below 4 mb/d by the December STEO.
B — The Impaired Grind · $75 to under $95 · 40% · base case
No settlement and no catastrophe. Transit grinds from 8 toward 10–11 mb/d as dark-fleet routing, ship-to-ship transfer off Oman and convoying mature into routine. Shut-ins decline slowly rather than snapping back. The war-risk premium stays embedded but stops widening; the freight market keeps absorbing the stress that the crude market does not. Prices drift lower with the arriving surplus without ever getting a reopening. This is where the EIA's $83, Goldman's implied $89.70 and much of HSBC's base case live, and it is the modal outcome for the simple reason that six months of this war have already produced exactly it.
What puts it in play: the status quo, extended. Tripwire: transit between 8 and 12 mb/d and monthly Brent inside $75–$95 for three consecutive months. What falsifies it: either tripwire above or below firing.
C — The Toll Regime · $95 to under $125 · 30%
The two paths into this bucket are different in character and we weight them together because they resolve identically.
The first is the metered reopening. The corridor is formalised with a charge. The 5–7%-of-cargo-value proposal reported to Reuters would be $4.25–$5.95 a barrel at $85 oil; the flat-fee variants are reported at around $1 a barrel, equivalently $1–2 million per vessel per voyage. A toll does two things at once: it releases barrels, which is disinflationary, and it installs a permanent wedge in the cost of moving a quarter of the world's seaborne oil, which is not. Our simulation puts the probability that a toll is in force by March 2027 at 28.5%, with a mean charge of $3.51 a barrel among the paths where one is levied.
The second is the grind that does not end. No settlement, and the slow attrition of bypass capacity — Bab el-Mandeb pressure on Yanbu, Suez rerouting, the mid-2027 UAE pipeline arriving too late — means transit stalls near 8 mb/d instead of improving. Morgan Stanley's $95 for Q1 2027 and HSBC's stalemate case near $120 are both in this territory.
What puts it in play: a published transit-fee schedule, or transit failing to exceed 9 mb/d by the year-end. Tripwire: any officially announced Hormuz transit charge, from any party, at any level. What falsifies it: transit above 14 mb/d with no fee.
D — The Break · $125 or higher · 5%
Kharg is permanently destroyed, or a sustained mining campaign closes the Strait outright, or the IRGC's threat against tankers at Kuwaiti and Bahraini piers is executed and Gulf loading itself stops, or a Saudi or Emirati export facility is lost.
We want to be explicit about why this weight is low, because it is the number most likely to be argued with. The market has already shown us what a near-closed Strait prices at. In the second quarter of 2026 Hormuz liquids transit ran at 4.9 mb/d — roughly a quarter of baseline — and Brent averaged $102.93 for the quarter, peaking at a monthly average of $117.29 in April. A monthly average at or above $125 is therefore above anything this war has produced at its worst. Our simulated escalation regime, which drives transit to 3 mb/d, prices at roughly $105–$108. Getting to $125 as a monthly average requires something outside the observed range: the loss of Gulf loading infrastructure rather than the loss of Gulf transit. Iran's own exports, at 135,000 b/d in August against 1.98 mb/d in February, are already almost entirely gone — there is very little Iranian supply left to remove.
Tripwire: confirmed mining inside the Strait, or a confirmed strike on Ras Tanura, Ras Laffan or Fujairah. What falsifies it: it is the tail; the base case falsifies it daily.
Simulation results
mc.py ships with this report. 50,000 paths, seed 20260909, numpy 2.2.6, seven monthly steps from September 2026 to March 2027. This is a real simulation and these are results, not a design.
The model does not forecast oil. It forecasts the Strait, and prices the consequence — because the $22 spread between named houses is a disagreement about the transit-recovery path and nothing else.
A rejected specification, documented rather than hidden. We first tested the obvious mapping: log Brent regressed on the contemporaneous world net-inventory-withdrawal balance, EIA STEO T3_STCHANGE_WORLD against BREPUUS, quarterly, 2000-Q1 to 2025-Q4, n = 104. It returns R² = 0.001 and a coefficient of the wrong sign. Balances do not price oil in levels. We discarded it and say so.
Structure. Monthly effective Hormuz liquids transit follows a three-regime Markov chain — Impaired (the current state, 8.0 mb/d), Reopening, Escalation — with monthly transition hazards. Price is built as a fundamental plus a premium that is a power function of the transit shortfall, plus a toll where one has been levied, times a Student-t(5) monthly shock.
The premium function is fitted, not assumed: premium($) = 0.9618 × shortfall^1.3474, estimated by least squares in logs on three sourced (transit, price) pairs — Q2 2026 at 4.9 mb/d and $102.93; August 2026 at 8.0 mb/d and $91.08; 2 July 2026 with the MoU in force at approximately baseline transit and $69.
Key parameters. Baseline transit 19.87 mb/d (IEA, FY2025). Fundamental no-disruption Brent glides $65 → $62, anchored on the pre-war January 2026 average of $66.60 and the EIA's own $63.94 for Q4 2027 against a 6.25 mb/d build. Monthly hazards: reopening 12%, escalation 7%, relapse 5%, de-escalation 20%. P(toll | reopening) = 55%, toll uniform on $1–$6. These hazards are Desk judgement, declared as such, and every one of them is moved in the envelope below.
Results.
| A <$75 | B $75–95 | C $95–125 | D ≥$125 | |
|---|---|---|---|---|
| Simulation | 33.82 | 43.43 | 21.57 | 1.19 |
Median $81.17. Mean $83.61. Standard deviation $16.05. P5/P10/P25/P50/P75/P90/P95 = $62.1 / $65.6 / $71.9 / $81.2 / $93.6 / $105.4 / $112.0. P(≥ $100) 15.8%. P(touching $125 in any month) 7.5%. P(a reopening is reached by March) 51.8%. P(a toll in force) 28.5%, mean $3.51/bbl among tolled paths.
Dispersion, measured rather than asserted. The achieved terminal log-change standard deviation is 0.1853 against an empirical unconditional seven-month figure of 0.2771 — the simulation is 33% under-dispersed. We report this rather than tuning it away. Part of the gap is legitimate: the unconditional history includes regime shifts that this model represents explicitly, and adding the full historical variance on top would double-count them. Part is not, and a reader who thinks the tails are too thin is entitled to that view. The wide-noise specification (σ = 0.189, the war-period figure) is in the envelope and takes D to 3.9%.
Seed stability across four seeds, bucket C: 21.57 / 21.72 / 21.46 / 22.11. Sampling error is negligible. Specification envelope, 25 specifications, same code and seed, one parameter moved at a time:
- A [18.2, 50.5] · B [33.1, 50.0] · C [4.5, 30.7] · D [0.2, 16.1]
The structural envelope is roughly fifty times the sampling envelope, and one parameter dominates it. Moving γ, the convexity exponent of the premium function, by ±15% takes bucket D from 0.2% to 16.1% and bucket C from 4.5% to 27.8%. Nothing else in the model comes close — not the escalation hazard, not the toll probability, not the noise. γ is fitted to three points. That is the single weakest joint in this report and the red team should start there.
The second most powerful parameter is the fundamental level F: ±$6 moves bucket A from 18.2% to 49.8%. Where oil settles once the Strait is open matters more than most of the war parameters.
MiroFish was not run and is not implied.
Aggregation
Standing Desk rule: base rate 0.30, expert priors 0.30, simulation 0.40.
| Leg | A | B | C | D |
|---|---|---|---|---|
Base rate (39 years, FRED MCOILBRENTEU) |
15.78 | 36.83 | 40.51 | 6.88 |
| Expert priors (five named, dated houses) | 22.79 | 44.89 | 31.51 | 0.81 |
| Simulation (50,000 paths) | 33.82 | 43.43 | 21.57 | 1.19 |
| Aggregate | 25.10 | 41.89 | 30.23 | 2.78 |
| Published (5% grid) | 25 | 40 | 30 | 5 |
The rounding took bucket D from 2.78 to 5 and bucket B from 41.89 to 40. The rounding rounded the tail up, which is the conservative direction; it is disclosed here rather than buried. The grid summed to 100 with no carry required.
The expert leg required a mapping the houses did not make, and it is ours. Four of the five published forecasts are 2027 annual averages; the resolution variable is the March monthly average. The EIA's own STEO puts March 2027 at $83 against a 2027 annual average of $74 — a ratio of 1.122 — and we applied that ratio to the annual forecasts. This inflates Citi's $65 to an implied $72.90 and HSBC's $85 to $95.30, which pushes HSBC just across a bucket boundary. The mapping is declared as a named source of error.
| House | Published | Date | Implied March 2027 |
|---|---|---|---|
| Citi | $65, 2027 average | 3 Sep 2026 | $72.90 |
| EIA STEO | $83, March 2027 exactly | 9 Sep 2026 | $83.00 |
| Goldman Sachs | $80, 2027 average | 7 Sep 2026 | $89.70 |
| Morgan Stanley | $95, Q1 2027 | note 23 Aug 2026 | $95.00 |
| HSBC (Kim Fustier) | $85, 2027 average | note 8 Sep 2026 | $95.30 |
Note the structure of the disagreement. The three legs do not merely differ in magnitude, they point in different directions on the bucket that matters. History says C at 40.5% — a six-month run of this size has historically stayed elevated. The structural model says C at 21.6% — the Strait probably reopens within seven months. The published 30% splits them, and we would rather show that than manufacture an agreement.
Market pricing versus the Desk view
What the market is pricing. The EIA's path — the closest thing to an official consensus — is a smooth monotone glide: $93, $92, $91, $89, $87, $85, $83, then $80, $77, $74 and on down to $62 by December 2027. Fifteen consecutive monthly declines, ten of them exactly $2. That path contains no branching whatsoever. It cannot, by construction, represent the possibility that the Strait reopens in November or does not reopen at all, and those are the only two things that matter.
Where we differ.
- The upper half is underpriced. We put 35% at or above $95. A point forecast of $83 puts 0% anywhere. The nearest thing to a published distribution is HSBC's, which spans $70s to $120 across its scenarios — a range wider than the entire five-house consensus band, published by the house closest to the shipping data.
- The toll is not in any price we can find. A $3.50-a-barrel permanent transit charge is levied on a waterway that has been free since the trade existed. At full pre-war throughput of 19.87 mb/d that is on the order of $25 billion a year of rent; the tolled paths in our own simulation carry a mean terminal transit of 10.73 mb/d, so the near-term figure is roughly half that, and it grows as traffic normalises. We can find no forecast that separates a free reopening from a metered one. They resolve at very different prices and, more importantly, they have very different half-lives: a war premium decays, a toll does not.
- We are not contrarian on the middle. Our median of $81 sits $2 below the EIA's $83. Anyone reading this expecting a dramatic call against the consensus should note that the central estimates agree. The disagreement is entirely about the width.
Highest-conviction mispricing: the distinction between a free reopening and a metered one, which the market currently treats as the same event.
What would prove the Desk wrong: Hormuz transit above 14 mb/d on a sustained basis before December with no fee of any kind announced, and shut-ins below 3 mb/d in the December STEO. That is bucket A, it carries 25%, and it is a perfectly ordinary outcome.
The universal-owner portfolio map
Directional, strategic, and not personalised advice. Magnitude bands are the Desk's, not modelled outputs.
| Asset class | A · Reopening | B · Grind | C · Toll Regime | D · Break |
|---|---|---|---|---|
| Global equities | + small | flat | − small | − large |
| Energy majors | − medium | + small | + medium | + large |
| Oil services / E&P | − medium | flat | + medium | + large |
| Tanker & LNG shipping | − large | + large | + large | + large |
| Airlines, road transport | + medium | flat | − medium | − large |
| Chemicals, fertiliser | + medium | − small | − medium | − large |
| Utilities (gas-exposed) | + medium | − small | − medium | − large |
| Long-duration sovereigns | + small | flat | − small | − medium |
| Inflation-linked bonds | − medium | flat | + medium | + large |
| Investment-grade credit | + small | flat | − small | − medium |
| Gulf sovereign credit | + small | flat | − small | − medium |
| EM oil importers (FX + local rates) | + large | − small | − large | − large |
| Gulf sovereign equity & SWF NAV | − small | + small | + medium | + medium |
| Marine insurance & reinsurance | − large | + medium | + medium | mixed: rate up, loss up |
| Infrastructure (ports, pipelines, storage) | flat | + medium | + large | + large |
| Defence and dual-use | − small | + small | + medium | + large |
| Gold & reserve assets | − small | flat | + small | + large |
Three observations a diversified owner should take from that table rather than from any single row.
You are long the reopening whether you chose to be or not. A universal owner holds the airlines, the chemicals, the EM importers and the long-duration bonds that all improve in bucket A, and holds them in far greater size than the energy and shipping exposures that improve in C. The portfolio has a directional view on the Strait embedded in it. Twenty-five percent is a large probability to be carrying passively on both sides of a single waterway.
Bucket C is the awkward one, and it is the one with 30%. It is not a crisis. It is a mild, permanent tax that shows up as a persistent inflation floor, a structurally higher cost of moving energy, and a windfall to owners of physical infrastructure outside the chokepoint. It is precisely the kind of outcome that a scenario framework built around "war or peace" cannot see, because it is both.
The real-asset leg is where a toll regime is expressible. If a charge is instituted, the durable beneficiaries are bypass pipelines, non-Gulf LNG capacity, storage, and ports outside the Strait — assets that universal owners buy directly and hold for decades, on horizons long enough for a structural change in chokepoint economics to be the whole investment case.
Second- and third-order effects
A toll is a precedent, not a transaction. If a mandatory charge for transit passage is established at Hormuz and survives, it is the first successful monetisation of an international strait in the modern era. UNCLOS Article 26 permits charges only as payment for specific services actually rendered, and transit passage "shall not be impeded" — but neither Iran nor the United States has ratified UNCLOS, and the Institute for the Study of War's Critical Threats Project reads the Iranian scheme as an attempt to normalise de facto control. Malacca, Bab el-Mandeb, the Turkish Straits, Panama and the emerging Arctic routes all have interested coastal states watching whether this works. For an owner of global shipping, port and trade-linked infrastructure on a thirty-year horizon, the precedent is worth more than the barrel.
The insurance market is doing the pricing the crude market is not. War-risk cover for Gulf transits was quoted at 7.5–10% of hull value in July against 1–3% weeks earlier, per Marsh's Marcus Baker — on a $100 million tanker that is $7.5–10 million a voyage against roughly $250,000 before the war. The Lloyd's Joint War Committee added Bahrain, Djibouti, Kuwait, Oman and Qatar to its listed areas in March. Insurance capacity, not naval escort, is the binding constraint on how fast transit can recover in a reopening. We could find no war-risk quote dated later than 22 July and mark this section accordingly.
Energy security is being re-domesticated, and that is a capex cycle. The bypass build-out is already visible: the UAE's new capacity arrives mid-2027, per the EIA. Every month the Strait stays impaired adds to a pipeline, storage and non-Gulf LNG investment programme with a twenty-year life, funded substantially by the same asset owners reading this.
The IEA's shock absorber has been used. Four hundred million barrels released in March was the largest collective action in the agency's history and the sixth ever. That buffer is not available again at the same scale in the same year. A tail event arriving in 2027 meets a thinner cushion than the same event arriving in February 2026 did — which argues, mildly, for a fatter D than the price history alone implies.
LNG's substitution has been quiet and remarkable. Non-Gulf output rose about 18% over March–June, offsetting roughly three-quarters of the Gulf shortfall; global LNG production fell only about 4% despite Qatar losing 96% of its cargoes. That elasticity is the strongest single argument for bucket A, and it deserves more attention than it is getting.
What we are watching
| # | Indicator | Latest reading | Threshold that moves the model |
|---|---|---|---|
| 1 | Hormuz liquids transit, Vortexa 28-day average | ~8.0 mb/d (31 Aug) | >14 → A live · <6 sustained → C live |
| 2 | Any announced transit fee, from any party | none | any published schedule → C live |
| 3 | EIA STEO shut-in production | 6.7 mb/d (Aug); 5.7 forecast Q4 | <4 by Dec → A · >8 → C |
| 4 | EIA STEO Q1-2027 inventory balance | −3.01 mb/d (a build) | revised to a draw → A dead |
| 5 | Kharg loadings, Kpler | 135 kb/d (Aug), from 1.98 mb/d (Feb) | any restart → A · terminal destruction → D |
| 6 | Yanbu loadings / Bab el-Mandeb security | −~50% m/m (Aug) | further halving → C |
| 7 | Iran–Oman corridor status | agreed 26 Aug, not operating | formalised with fee → C · without → A |
| 8 | Lloyd's JWC listed areas & war-risk rate | 7.5–10% of hull (22 Jul) | <3% → A · >15% → C/D |
| 9 | Baltic TD3C / TD34 | WS677 / WS272 (4 Sep) | TD34 <WS150 → A |
| 10 | Bahrain 5-year CDS | 284 bp, +101 YTD | >400 → systemic stress |
| 11 | Saudi 5-year CDS | −9 bp YTD | >+50 → D live |
| 12 | Qatar LNG cargo count | 18 in six months vs 509 | >60/month → A |
| 13 | OPEC+ decisions | +188 kb/d for Sept, agreed 2 Aug | a cut → the surplus thesis weakens |
| 14 | Brent monthly average, FRED MCOILBRENTEU |
$91.08 (Aug) | three months >$95 → C · <$75 → A |
| 15 | US–Iran talks / a successor to the June MoU | lapsed | signature → A |
Interim checkpoints: 7 October and 5 November 2026 (STEO), December 2026 STEO for the shut-in path, and the March 2027 OPEC and IEA reports. Resolved on publication of the March 2027 MCOILBRENTEU value, expected early April 2027.
Red team — how this could be wrong
1. γ is fitted to three points and it dominates everything. The premium function's convexity exponent moves bucket D from 0.2% to 16.1% across a ±15% range, and there are three observations behind it. If the true relationship is more convex than 1.35 — which is exactly what one would expect if the remaining transit is progressively harder to replace, because the easy substitutions get made first — then our D of 5% is far too low and the honest number could be in the teens. This is the report's weakest joint. We have published the parameter, the fit, the data behind it and the sensitivity so a reader can disagree quantitatively rather than rhetorically.
2. We may be systematically too willing to believe in reopening. Our hazards give a 51.8% chance of reaching a reopening within seven months. That is a judgement, not a measurement. The June MoU is the only evidence we have about how these settlements go, and it collapsed in three weeks. A reader who sets the reopening hazard at 6% a month gets bucket A at 23.6% and C at 26.8% — a materially different report from the same code.
3. The base rate may be the wrong reference class entirely. Seven usable analogue windows is a very small sample, we discarded one of eight for contamination, and the two closest structural analogues to a metered chokepoint — the 1973 and 1979 episodes — resemble the current case in the one dimension that matters (an administered rather than a cleared price) while differing in every other. If the toll is instituted, the right prior may be the 1970s rather than the 1990s, and the 1970s prior is much higher.
4. The under-dispersion is real. A 33% shortfall against the historical seven-month standard deviation is not nothing. Both tails are probably too thin, and the fix would raise A and D together at the expense of B and C.
5. The data is genuinely contested and we picked a side. Transit estimates for the same week range from 5.94 mb/d (TankerTrackers) to 18 mb/d (the President of the United States). We used the commercial trackers, which is defensible and is also a choice. If the official figures are closer to right, the shortfall driving our premium function is overstated and the whole distribution shifts down.
6. What could make this wrong tomorrow. A signed successor to the June memorandum would put bucket A in play within days. Confirmed mining inside the Strait, or a strike on Ras Tanura or Ras Laffan, would put D in play within hours. Both are ordinary events in a war that has been running for six months.
Methodology
Three legs, aggregated 0.30 base rate / 0.30 expert priors / 0.40 simulation. The base-rate leg is an equal-weight blend of four reference classes computed on FRED MCOILBRENTEU, May 1987 to August 2026, n = 472, in base_rate.py, which ships with this report and runs offline from its own data slice. The expert leg is a five-component normal mixture over the implied March 2027 values of five named, dated published forecasts, standard deviation $14, which is Desk judgement anchored on the only house to publish a scenario range. The simulation is mc.py: 50,000 paths, seed 20260909, numpy 2.2.6, a three-regime Markov chain on Hormuz transit mapped to price through a fitted power-law premium function, with Student-t(5) monthly shocks. aggregate.py reproduces the published weights from the three legs' outputs. Every figure in this report is reproducible from the four scripts and three data files in the accompanying package. Live macro and energy data via research/data_spine.py (FRED, EIA). One specification was tested and rejected and is documented in mc.py. MiroFish was not run and is not implied. Weights sum to 100 and are rounded to 5%; the rounding is disclosed above.
Editorial scenario analysis only. Not investment, actuarial, or geopolitical advice. This report is for informational and research purposes only and does not constitute investment, legal, tax, or financial advice.
Source ledger
Every figure used in this report, with the source that carries it and the date it was published. Confidence is H/M/L.
| # | Source | Date | Data point used | Conf |
|---|---|---|---|---|
| 1 | U.S. Central Command, public release; Stars and Stripes (Bloomberg/TNS) | 8–9 Sep 2026 | Five IRGC tankers destroyed 8 Sep; three more on 5 Sep; no US injuries; Rubio quote | H |
| 2 | Stars and Stripes, 9 Sep 2026 | 9 Sep 2026 | Brent settled ~$98; Kharg explosions; Iran missiles at Jordan; IRGC warning re Kuwaiti/Bahraini piers; Bessent quote; one-fifth of world oil and LNG pre-war | H |
| 3 | FRED MCOILBRENTEU (EIA source data) |
pulled 9 Sep 2026 | Brent monthly averages 1987-05 to 2026-08; Aug 2026 $91.08; Apr 2026 peak $117.29; Jan 2026 $66.60 | H |
| 4 | FRED DCOILBRENTEU |
pulled 9 Sep 2026 | Daily Brent; last print $96.02 (1 Sep 2026) | H |
| 5 | EIA Short-Term Energy Outlook, series BREPUUS |
released 9 Sep 2026 | Brent monthly path: Mar-2027 $83; Sep-2026 $93 through Dec-2027 $62 | H |
| 6 | EIA STEO, series T3_STCHANGE_WORLD |
released 9 Sep 2026 | World net inventory withdrawals: Q4-26 +1.71 mb/d → Q1-27 −3.01 mb/d | H |
| 7 | EIA STEO, global oil report | released 9 Sep 2026 | Shut-ins 6.7 mb/d Aug vs 5.0 Jul; 5.7 forecast Q4; Yanbu loadings −~50% m/m; new UAE bypass mid-2027; Brent Aug avg $91 | H |
| 8 | EIA STEO (August 2026 edition) | 11 Aug 2026 | Brent $69 on 2 Jul 2026; $105 on 23 Jul 2026; Hormuz 4.9 mb/d in Q2-26 vs 21.6 mb/d in Q4-25 | H |
| 9 | IEA, Strait of Hormuz factsheet (IEA analysis of Kpler data) | last updated Feb 2026 | FY2025 transit 19.87 mb/d; LNG 112 bcm ≈ 20% of global trade; bypass capacity 3.5–5.5 mb/d; Petroline and ADCOP specifications | H |
| 10 | Vortexa (Rohit Rathod), via Axios | 4 Sep 2026 | Hormuz ~8 mb/d, 28-day moving average to 31 Aug 2026 | H |
| 11 | TankerTrackers, via Axios | 4 Sep 2026 | Crude specifically 5.94 mb/d, 7-day average | H |
| 12 | Kpler and Lloyd's List Intelligence (Bridget Diakun), via Al Jazeera | 3 Sep 2026 | 6/11/5 daily transits; 10-day average 13 vessels; ~12 transits/day 26 Aug–1 Sep; dark-transit caveat | H |
| 13 | Drewry (Eirik Hooper), via Al Jazeera | 3 Sep 2026 | Reconciliation of official vs commercial transit counts | H |
| 14 | U.S. Department of Energy spokesperson; Energy Secretary Chris Wright (CNBC); President Trump (Truth Social) | 2–4 Sep 2026 | ~12 mb/d on 2 Sep; "17 million barrels" on 31 Aug; "18 million barrels per day" | M |
| 15 | HSBC (Kim Fustier), research note | 8 Sep 2026 | 2027 Brent $85, from $65; 2026 $90; flows "~30% of pre-conflict"; 6→8→9.5 mb/d path; stalemate ~$120; recovery $70s by 1Q28 | H |
| 16 | Morgan Stanley, research note (via Reuters) | note 23 Aug 2026 | Q1-2027 Brent $95; Q2-27 $90; market in deficit through Q4 and Q1 | H |
| 17 | Goldman Sachs | 7 Sep 2026 | 2027 Brent $80; Dec-2026 $85; both raised $5 | H |
| 18 | Citi, research note | 3 Sep 2026 | 2027 Brent $65 unchanged; Q3-26 raised to $86; larger-than-pre-conflict surplus on reopening | M |
| 19 | Al Jazeera | 26 Aug 2026 | Iran–Oman corridor: 7 miles wide, Iranian waters, Iranian permission; Gharibabadi, Akraminia, Albusaidi named; IRGC "share… in its revenues"; preconditions tied to the 17 Jun MoU; pre-war Brent ~$66; Mar-2026 peak ~$120 | H |
| 20 | Reuters (senior Iranian official), via Gulf News | 7 Aug 2026 | Transit fee proposal of 5–7% of cargo value; flat-fee alternatives ~$1–2/bbl or ~$2m per VLCC; Oman's Malacca-style voluntary counter-proposal; UNCLOS Art. 26; ISW Critical Threats reading | M |
| 21 | Kpler, via The National (Fareed Rahman) | 31 Aug 2026 | Kharg loadings 1.98 mb/d (Feb) → 135,000 b/d (Aug); Kharg ≈90% of Iranian exports; 7 mb/d loading capacity; 30 mb storage | H |
| 22 | Rystad Energy (Lin Ye), via The National | 31 Aug 2026 | Larak Island strike had "limited direct impact on Iran's oil exports" | H |
| 23 | Baltic Exchange, Tanker Report Week 36 | 4 Sep 2026 | TD3C WS677.22, TCE ~$704,000/day; TD34 WS272.5, ~$261,800/day | H |
| 24 | Baltic Exchange, Tanker Report Week 4 | 23 Jan 2026 | TD3C WS126.5, TCE $112,394/day — the pre-war baseline | H |
| 25 | Lloyd's List | 2026, undated on page | Caveat that TD3C is assessed on a route many owners will not sail | M |
| 26 | Marsh (Marcus Baker), via S&P Global Commodity Insights | 22 Jul 2026 | War-risk premium 7.5–10% of hull value, from 1–3% weeks earlier | H |
| 27 | Lloyd's Market Association Joint War Committee, circular JWLA-033 | 3 Mar 2026 | Bahrain, Djibouti, Kuwait, Oman, Qatar added to listed areas | M |
| 28 | The National | 17 Jul 2026 | War-risk 3–10% of hull vs ~0.25% pre-war | M |
| 29 | AGBI (Matt Smith), citing Arqaam Capital (Fady Gendy) and UBP (Apostolos Bantis) | 28 Aug 2026 | GCC CDS YTD to 24 Aug: Saudi −9, UAE +6, Qatar +6, Kuwait +4, Oman +7, Bahrain +101 at 284 bp; Qatar SWF+reserves 300% of GDP, Kuwait 606% | H |
| 30 | Euronews, citing Kpler, Vortexa and the IEA Gas Market Report Q3 2026 | 8 Sep 2026 | Qatar 18 LNG cargoes in six months vs 509; ~$24bn lost sales; Ras Laffan 140–150/month pre-war; non-Gulf LNG +18%; global LNG −4%; QatarEnergy force majeure into November; three STS transfers in August | H |
| 31 | IEA, collective action update | 15 Mar 2026 | 400 mb released — largest in IEA history, sixth ever; 271.7 mb government, 116.6 mb industry; US 172.2 mb | H |
| 32 | IEA Oil Market Report, August 2026 | 12 Aug 2026 | 2027 supply exceeds demand by >5 mb/d; 2026 supply −3.9 mb/d; observed inventories below 7.9bn bbl in July | M |
| 33 | OPEC press release / OPEC+ | 2 Aug 2026 | +188 kb/d for September; completes reversal of ~3.5 mb/d of 2023 cuts; quotas expected flat for the rest of 2026 | M |
| 34 | OPEC Monthly Oil Market Report | 12 Aug 2026 | 2027 demand growth revised up to ~2.2 mb/d y/y | M |
| 35 | ECB Economic Bulletin 8/2023 — Ferrari Minesso, Lappe, Rößler, "Geopolitical risk and oil prices" | Dec 2023 | Geopolitical oil pressures "insignificant after one quarter"; Ukraine +30% in two weeks, retraced in ~8 weeks; Russia the +2% exception via the embargo channel | H |
| 36 | Caldara & Iacoviello, "Measuring Geopolitical Risk," American Economic Review 112(4) | 2022 | The GPR index underlying the ECB result | H |
| 37 | Dallas Fed Working Paper 2403 — Kilian, Plante, Richter | 3 Jun 2024, rev. 18 May 2025 | A 20pp rise in the probability of a 5% shortfall reduces output 0.12% | H |
| 38 | Dallas Fed — Plante & Richter, "How Times Have Changed" | 23 Jun 2026 | US GDP response to the 2026 Iran war ~1/20th of a 1980 equivalent; ~15% global supply disruption | M |
| 39 | Strauss Center (UT Austin), citing Navias & Hooton, Tanker Wars (1996) p.183, and Blair & Lieberthal, Foreign Affairs 86:3 (2007) | accessed 9 Sep 2026 | 239 tankers attacked 1984–88, 55 sunk or written off, 61% of all ships attacked; never more than 2% of transits disrupted; Strait never closed | H |
| 40 | EIA Today in Energy #41413 | Sep 2019 | Abqaiq: 5.7 mb/d disrupted; intraday decay $71.57 → $65.09 within nine hours; prices back to pre-attack by month-end | H |
| 41 | EIA Today in Energy #65884 | 2025 | Israel–Iran June 2025: Brent $69 → $79 12–19 Jun, quarter-end $68 | H |
| 42 | EIA Today in Energy #61363 | 2024 | Red Sea: MEG–Netherlands product tanker rates $23,000 → $73,000/day; Shanghai–Genoa box $1,400 → $6,300 | H |
| 43 | Oil & Gas 360; FRBSF Weekly Letter (6 Jun 1986) | 2016 / 1986 | OPEC basket $23.29 (Dec 1985) → $9.85 (Jul 1986); cause was Saudi abandonment of the swing-producer role | M |
| 44 | Federal Reserve History, "Oil Shock of 1973–74" | — | 1973 embargo: $2.90 → $11.65 by Jan 1974; structural conditions of the 1970s premia | M |
| 45 | Econlib (James Hamilton), "Who Caused the August 1990 Spike in Oil Prices?" | 2014 | 1990–91: spot ~$28 by 6 Aug, ~$40 mid-Oct, back to ~$20 as Desert Storm began | M |
| 46 | S&P Global; Fortune | 10 & 28 Mar 2026 | Saudi East–West pipeline pushed to a record 7 mb/d in Q1 2026 | M |
| 47 | Seatrade Maritime | 8 Sep 2026 | Five Iranian tankers destroyed near Kharg — corroboration of #1 | H |
| 48 | Al Jazeera; NPR | 9 Aug 2026 | Houthi drone attack on the 400,000 b/d Jazan refinery — an August event, dated here because it is being recirculated in September coverage as current | H |
Items we could not source and therefore did not use: a Brent forward curve or 12-month spread for 9 September 2026 (no dated futures strip located; one widely-served contract table proved stale to 6 February 2026 and was rejected); a war-risk premium quote dated later than 22 July 2026; a current 2026 OPEC spare-capacity estimate; the outcome of the OPEC+ meeting of 6 September 2026; current Petroline and ADCOP throughput; and a single reconcilable J.P. Morgan 2027 Brent number (three incompatible figures surfaced across secondary reports). Each of these is a gap in this report, not an inference.
The Probability Desk is the afternoon intelligence property of Universal Asset Owners. We publish the evidence that moved every probability, we keep a public calibration log, and we grade ourselves.
Editorial scenario analysis only. Not investment, actuarial, or geopolitical advice.
Continue the briefing. Read the daily brief · watch the daily video briefing · listen to The Universal Owner · view the chart of the day.
Produced and edited by the UAO editorial desk. Not investment advice.