The Public Ship Count Through Hormuz Reads 96 Percent Lower. Nobody Can Say How Much of That Is Real.

By Kresmion Research, August 13, 2026

The IMF's PortWatch project publishes a daily count of ships passing the Strait of Hormuz, derived from the AIS transponder signals vessels broadcast. In February 2026 it averaged 69.9 transit calls a day. In March it averaged 2.8, a 96 percent drop, and since March 1 it has printed 17 days of literal zero against none in the 286 tracked days before.

Two true stories collided in that number on the same week, and this note is about the fact that no public data source can take them apart again. The first story: Iran closed the strait to most commercial shipping on February 28, the day joint US and Israeli operations began, and a US Navy blockade followed in April. Traffic really collapsed. The second story: the same waters came under GPS interference at a scale that put false positions under more than 1,100 vessels in a single day, and ships in the area went dark in large numbers, a practice still being reported in August. The counting layer really broke.

Both are documented. Their sizes are not separable. We spent a day and three adversarial fact-check rounds trying to attribute the collapse to one of them, our first draft got the answer wrong, and the version you are reading is the one the evidence survived. The conclusion is narrower than the one we started with, and we think it matters more: for five months, the most watched shipping lane on earth has had no public number anyone should quote.

Key takeaways

The series

Monthly averages of daily AIS-visible transit calls for the Strait of Hormuz, every month in our stored copy of the IMF PortWatch series. The window opens 2025-05-19, so the first row is a partial month, as is August 2026.

The last normal reading is February 28, at 51 transits, the war's first day. The first broken one is March 1, at 18. The count then follows the conflict's calendar: an average of 3.1 through the April 7 to 8 ceasefire, 3.8 through the quiet weeks after Operation Epic Fury concluded on May 5, a climb to 18.4 while the June 14 truce memorandum held, with single days of 44 and 43, and back to 4.2 within days of the truce ending on July 7. The chronology is Britannica's; the transit figures are the IMF's, and we verified our copy against the IMF's raw API, which returns values identical to our stored rows.

The collapse is real

Our first draft of this note argued the opposite, so the evidence for a genuine collapse deserves to be listed plainly.

Iran closed the strait to most commercial shipping on February 28, per Geopera's analysis. Britannica records a US Navy blockade of the strait from April 13, reimposed on July 14, and reports that commercial traffic through the strait dropped more than 90 percent at the war's peak, with diversion rates rising above 95 percent and fuel shortages and rationing across Asia. The US Treasury issued a 30 day waiver on March 12 for purchases of Russian crude already loaded on tankers, per analysis by the National Council on US-Arab Relations, the kind of step a government takes when real barrels have stopped moving.

The most direct physical evidence is satellite based and does not depend on AIS at all. Geopera's imagery analysis puts supertanker crossings at well over a dozen a day before the crisis, about eight a day in late June while the truce held, and roughly two a day after the truce broke. Ships that do not broadcast still show up in pictures. The pictures say the collapse in actual crossings is large.

The oil market agrees. Brent's front month averaged 69.41 dollars across February's closes, then 99.60 in March, 102.46 in April, and 104.09 in May, a sustained premium of roughly 45 percent for three months, with a closing peak of 118.35 on March 31. It printed 100.69 as recently as July 23 and closed at 88.50 on August 12, still about 27 percent above the February average. The US Energy Information Administration calls Hormuz the world's most important oil transit chokepoint, with flows equal to about 20 percent of global petroleum liquids consumption. This is what pricing a real, ongoing disruption of that artery looks like.

The measurement is also broken

If the collapse is real, why not simply read the count as a measurement of it? Because the counting layer was attacked in the same week the traffic fell, and the distortion is documented at the level of individual ships.

Windward, a maritime intelligence firm, reported more than 1,100 vessels experiencing GPS and AIS interference across the Gulf region within 24 hours, with vessels falsely positioned at airports, at a nuclear power plant, and on land in Iran, Oman, and the United Arab Emirates. A transit count built on positions does not know what to do with a tanker that reports itself in a parking lot. Coverage in Scientific American describes captains inside the strait watching chart displays show hundreds of ships circling false locations, against normal physical traffic it puts at roughly 130 to 150 vessels a day.

The single most useful number for judging the count came on May 5: Windward counted 167 commercial-size vessels in the strait area with 146 of them, 87 percent, broadcasting no AIS position. Whatever those 167 hulls were doing, an AIS-based count could see at most 21 of them. A census that can observe 13 percent of its population is not a census, whichever direction the population is moving.

That snapshot is three months old, and the practice it captured has not ended. On August 8, TankerTrackers reported a fleet of at least 56 visually identified tankers operating in digital darkness to support oil exports from Arab nations neighboring Iran, in its feed carried on our OSINT lane. Dark sailing in the strait is a present-tense, visually confirmed fact, observed by trackers who cannot rely on the ships' own radios either.

The three public numbers also refuse to reconcile with each other, which is its own warning. PortWatch counts transit calls under its own method, about 70 a day in the clean baseline. Scientific American's figure is daily vessel throughput, 130 to 150. Windward's May 5 figure is a snapshot of presence in an area, 167. Geopera's satellite series counts supertanker crossings only, a dozen or more a day pre-crisis. None of these converts into another, and any sentence comparing the broken count to one of them silently changes the units. We wrote such sentences in our first draft.

As of mid-August, the closest thing to a current measurement is a public dispute carried in our OSINT feed: most tanker trackers put non-Iranian flows through the strait near 5 million barrels a day, while the US government claims almost 9 million on a seven day average. The two figures describe the same water and differ by nearly a factor of two. When the sensing layer under a market is broken, this is what the numbers built on it look like.

The two broken series are the two wars

PortWatch tracks 28 chokepoints with the same method, and the geography of the failures is the strongest structural evidence about what failed. Twenty six run normally through the entire window. Malacca held monthly averages between 216.7 and 242.5. Bab el-Mandeb, with its own history of attacks and rerouting, held between 30.9 and 37.6 in every full month, and sits at 25.1 over the nine days of August so far, soft but continuous.

The two exceptions are the Strait of Hormuz and the Kerch Strait. Kerch, at the edge of the Russia-Ukraine war, fell from a 15.9 average before March to 9.5 after, with 14 zero days since March against one before. On the same like-for-like basis, since-March average against prior average, Hormuz is at 8.2 percent of its prior level and Kerch at 60 percent. Both broken series sit inside active war zones, where real disruption and electronic interference arrive together. That is precisely why neither series can be read at face value: the mechanism that reduces the traffic and the mechanism that blinds the sensor share a cause and a timeline.

One alternative explanation stays open. A provider that detects spoofing might discard suspect tracks exactly where spoofing happens, so a regional processing change inside the IMF's pipeline could shape the Hormuz series too. From outside, a filter dropping spoofed tracks and ships going silent look identical. We cannot rule this in or out, and it does not change the conclusion; it is one more reason the number resists interpretation.

What we got wrong on the way here

This note went through three adversarial fact-check rounds. The first two each overturned its central claim; the third corrected dates and wording in the text you are reading. It seems more useful to say so than to pretend otherwise.

Our first draft concluded the collapse was mostly an artifact: the ships kept sailing and only the radios went dark. That reading leaned on the May 5 presence count, 167 vessels with 146 dark, and it quietly treated presence as transit. Ships waiting at anchor are present; they are not crossing. It also leaned on a claim that the count never recovered after the fighting stopped, which our own stored series contradicts: the count tripled while the June truce memorandum held. And it dismissed the oil market's verdict by quoting a single retraced close instead of three months of averages near 100 dollars.

The second fact-check round surfaced the satellite crossing counts and the blockade record, and the artifact-only reading did not survive them. What survived is smaller and harder: a real collapse and a broken sensor, entangled beyond public separation. Our editorial rule is that a number we cannot stand behind does not get published as a number. This piece exists because the rule fired on our own first take, twice.

What we show instead of the number

Kresmion's physical intelligence map now marks the Strait of Hormuz as AIS-dark since late February, with the cause stated, and withholds the transit count from the display. The label is built to remove itself: if the seven day average climbs back above 35 transit calls, half the last clean month, the number returns. June's partial recovery tested that threshold and stopped at a seven day average of 26.7, so the label held. Bab el-Mandeb, whose numbers stayed continuous and plausible, keeps them.

Anything downstream of AIS in the Gulf inherits this problem silently: transit tallies, port-call statistics, netflow estimates, congestion indices. The honest treatment we can offer is a labeled gap. A chart that keeps rendering 4 a day as if it were a measurement would be quoting the jamming with the IMF's logo on it, and a chart that shows nothing at all would hide that something historic is happening to the strait. The label carries both facts.

What would change the read

Three observable things would move this from entangled to attributable. Satellite crossing counts and the AIS count converging again, with the dark-ship share falling, would mark the sensor healing while traffic recovers, and our label lifts itself at a sustained seven day average above 35. The AIS count rising while satellite counts and dark-ship counts stay flat would point at a processing change inside the IMF's pipeline rather than recovery on the water. And a settlement that reopens the strait would give the first clean month since February, a new baseline against which the whole broken stretch can finally be sized.

Frequently asked questions

Is the Strait of Hormuz closed right now?

Neither fully closed nor fully open, by the most recent reporting. Geopera's July assessment was that it remained effectively closed to most commercial traffic, and Britannica records the naval blockade being reimposed on July 14. Kresmion's OSINT feed, which ingests maritime and geopolitical reporting continuously, carries 342 strait-related items over the past 30 days, and the freshest of them, August 10 to 13, describe a contested and partially open waterway: Iran and Oman reported in the final stage of reopening talks tied to US terms, the US president and Iran each publicly claiming control of the strait this week, a missile reported launched toward the strait on August 12, and energy columnist Javier Blas writing that the strait is neither fully closed nor fully open, with most tanker trackers putting non-Iranian flows near 5 million barrels a day while the US government claims almost 9 million, against a prewar load near 21 million barrels a day, the figure in the National Council on US-Arab Relations analysis cited below. Those are third-party reports, carried with their sources' voices. What none of them can do is turn the AIS transit count back into a measurement, which is this note's subject.

How much of the 96 percent drop is real?

It cannot be determined from public data. Satellite counts of supertanker crossings show a large real collapse, from over a dozen a day to single digits. The 87 percent dark share among ships present shows the count is also blind to most of what remains. Both effects are large, they overlap in time and place, and no public series separates them. Flow estimates exist for the barrels: most tanker trackers put non-Iranian flows near 5 million barrels a day, while the US government claims almost 9 million. They disagree by nearly a factor of two, and neither number maps onto the transit count.

Why write about a number you refuse to display?

Because the number is still being consumed. PortWatch feeds research, risk models, and news coverage, and a reader taking the Hormuz series at face value gets a figure whose relationship to actual shipping is unknowable. Saying clearly why we will not quote it seemed more useful than either quoting it or ignoring it.

What did the fact-check change?

The conclusion. The first draft said the collapse was mostly a measurement artifact and the ships kept sailing. Adversarial review against our own sources found satellite evidence of a real, large fall in crossings, a naval blockade we had not weighed, and a presence-versus-transit confusion in our reasoning. The published conclusion, that the real and electronic effects cannot be separated, is what survived.

How was this verified?

Every transit figure was re-derived from our stored copy of the IMF series, which we matched row for row against the IMF's raw API. Every external claim was checked against the cited page, across three review rounds, by a reviewer instructed to refute the piece. The war chronology is Britannica's. The vessel-presence and interference figures are Windward's. The satellite crossing counts are Geopera's. The price history is our own stored Brent series.

Sources

IMF PortWatch, daily chokepoint transit calls, Strait of Hormuz, Bab el-Mandeb, Malacca, Kerch. https://portwatch.imf.org/

Kresmion stored copy of the PortWatch series, 2025-05-19 to 2026-08-09, with raw upstream ArcGIS values matched against stored rows on 2026-08-13.

Kresmion futures price history, Brent front month closes, 2026-02-27 to 2026-08-12.

Windward, Widespread GPS Jamming Hits 1,000-plus Ships in the Middle East. https://windward.ai/blog/gps-jamming-disrupts-1100-ships-in-the-middle-east-gulf/

Geopera, Strait of Hormuz Satellite Imagery: Tracking Ships Gone Dark. https://geopera.com/blog/strait-of-hormuz-satellite-imagery

Scientific American, GPS spoofing is scrambling ships in the Strait of Hormuz. https://www.scientificamerican.com/article/gps-spoofing-is-scrambling-ships-in-the-strait-of-hormuz/

National Council on US-Arab Relations, Operation Epic Fury and Iran's Maritime War. https://ncusar.org/aa/2026/03/operation-epic-fury-and-irans-maritime-war-how-tehrans-defiance-is-reshaping-global-economic-security/

Britannica, 2026 Iran war, chronology, blockade, and traffic diversion. https://www.britannica.com/event/2026-Iran-war

US Energy Information Administration, World Oil Transit Chokepoints. https://www.eia.gov/international/analysis/special-topics/World_Oil_Transit_Chokepoints