Incident Counts Answer Only One Question
A confirmed attack count tells a company how many events met the reporting or validation threshold during a period. It does not directly tell the probability that the next ship will be attacked.
If 10 attacks occur among 1,000 transits, the exposure differs from 5 attacks among 20 transits. The second period has fewer attacks but a much higher crude incident-to-transit ratio. That ratio is still not a true probability because ships and events are not identical, yet it is more informative than the count alone.
Commercial updates should therefore avoid phrases such as "risk fell because attacks declined" unless traffic, geography, severity, and reporting conditions support the conclusion.
Traffic Data Are Not One Consistent Population
JMIC's 12 November assessment said the 2025 historical average was about 138 vessels per day. For 8 to 11 November, it reported 135 US-facilitated Hormuz transits from NCAGS over 96 hours. It separately referred to single-digit or low-double-digit AIS-visible transits in internal tracking.
Those figures are not necessarily contradictory. The advisory explained that NCAGS data covered US-facilitated transits, while S&P Maritime and SeaVision metrics covered non-facilitated cargo-carrying ships above a stated size and excluded some Iranian-port traffic.
Adding, comparing, or dividing such figures without understanding overlap and exclusions can create a false result. Analysts should record source, coverage, vessel threshold, geography, direction, AIS requirement, and time basis for every traffic number.
AIS-Dark and Spoofed Ships Complicate the Denominator
AIS-derived traffic misses ships that are not transmitting, are outside receiver coverage, are filtered by size or type, or are represented by corrupted data. In the Gulf, JMIC repeatedly reported GNSS interference, AIS anomalies, spoofing, and jamming.
A low visible-transit count can therefore mean few ships sailed, many ships sailed dark, tracking failed, or the dataset excludes part of the fleet. The likely explanation may be a combination.
Companies should not "correct" the count with an invented multiplier. They should show a range or state that exposure is uncertain. Where possible, AIS should be cross-checked against reporting-center data, port movements, satellite information, agents, and coastal-state sources.
Reporting Delay Changes the Numerator
Incidents can be reported late, reclassified, duplicated, or removed when better evidence arrives. UKMTO's public incident records include time-late reports and updates that change an event's classification.
A quiet 24-hour window may later contain an attack once the flag state, company, or authority confirms it. Comparing today's provisional count with last month's mature count biases the conclusion.
Dashboards should display incident date separately from report date and mark the data cutoff. Analysts should use a fixed maturation window when comparing periods or explicitly state that the recent period is incomplete.
Severity Must Sit Beside Frequency
Five incidents with no injury or loss of propulsion are not commercially equivalent to one fatal strike that sinks a ship, causes pollution, or closes a route.
A risk table should track fatalities, injuries, abandonment, sinking, propulsion loss, fire, pollution, detention, and repeat strikes. It should also distinguish deliberate attack, collateral damage, suspicious activity, and navigation casualty.
The commercial response often follows the worst plausible consequence rather than the average event. Insurers and crews may react sharply to a single fatal attack even when total incident frequency declines.
Geography and Behavior Change Exposure
A transit through a coordinated corridor at a selected time is not equivalent to drifting near a recent attack, waiting at a congested anchorage, or operating close to military units.
The denominator should ideally be exposure units relevant to the hazard: transits through a defined zone, vessel-hours at anchor, calls at affected terminals, or movements under a particular framework. Such data are rarely complete in real time.
A crude attack-per-transit ratio can still support directional analysis, but it should not erase differences in route, ship type, affiliation, speed, AIS policy, naval assistance, and time spent exposed.
Avoid the Survivor-Bias Trap
Analysts often study ships that sailed and returned safely. They may overlook ships that cancelled, waited, diverted, or were screened out by owners and insurers.
If only the most prepared ships, strongest operators, or accepted assisted transits proceed, their outcomes do not prove that the route is safe for all ships. The observed fleet has been selected.
Conversely, ships that proceed despite high risk may differ in ways that increase exposure. Without knowing the selection process, a successful-transit percentage should not be generalized.
Decision-makers should ask who did not sail and why.
A Practical Exposure Dashboard
ZeaClub recommends five panels.
Traffic: transits by direction, source, coverage, and confidence.
Incidents: event and report dates, classification, location, ship type, and verification status.
Consequences: casualties, damage, abandonment, pollution, delay, and route effect.
Stationary exposure: vessels at anchor, berth, drifting, disabled, or waiting.
Controls: assisted movements, route changes, mine clearance, threat levels, insurance availability, and reporting quality.
The dashboard should show rates only where numerator and denominator cover comparable geography, time, and vessel populations. Otherwise, the data should remain adjacent without a misleading division.
What Can Be Calculated Responsibly?
A crude exposure indicator can be calculated as verified incidents divided by observed relevant transits for the same zone and period. It should be labeled an indicator, not an individual ship's probability of attack.
A range can reflect uncertain traffic coverage. Analysts can also track attacks per week alongside traffic as an index, severity-weighted incidents, or vessel-hours if reliable data exist.
They should not assign arbitrary severity weights and present the result as fact. Any weighting is a model choice. The source data, assumptions, and sensitivity should be visible.
A Hypothetical Weekly Comparison
Assume Week A has six verified attacks and 120 observed transits. Week B has three attacks and 20 observed transits. The headline count halves, but the crude attack-to-observed-transit indicator rises from 5 percent to 15 percent.
Those figures do not prove that every Week B transit had a 15 percent attack probability. Ships may differ, reports may be incomplete, AIS-dark traffic may be missing, and one incident may involve a non-transiting vessel.
They do prove that "attacks fell by half" is not a sufficient safety conclusion. The company should investigate traffic contraction, route, ship mix, controls, and incident severity before changing posture.
Market Prices Are Another Imperfect Denominator
UNCTAD reported sharp increases in tanker shipping costs during the Hormuz disruption and attributed pressure to reduced tanker availability, time-consuming transfers, and longer routes. Freight and war-risk prices can reveal scarcity and perceived danger.
They do not independently verify security improvement or deterioration. Rates also respond to cargo demand, fleet positioning, fuel, congestion, and contract timing.
A falling premium may reflect new capacity rather than lower physical risk. A rising rate may reflect fewer willing owners even if attacks are unchanged. Market data should corroborate, not replace, operational evidence.
De-escalation Requires More Than a Quiet Interval
Meaningful de-escalation may include sustained reduction in severe verified incidents, restored and repeatable traffic, lower stationary backlogs, cleared hazards, reliable navigation, reduced official threat levels, and workable insurance.
The duration matters. A single quiet period in a low-traffic environment provides weak evidence. Several weeks of safe operations across a broader ship population provide more, although never absolute proof.
JMIC explicitly warned earlier in the conflict that fluctuations in reported strikes did not necessarily represent meaningful de-escalation. That principle remains sound because intent, capability, and exposure can persist beneath a quiet count.
The ZeaClub View
The most dangerous maritime statistic is a numerator without its denominator. The second most dangerous is a denominator whose coverage no one checked.
Fewer attacks can be good news, but only when the fall is interpreted alongside traffic, reporting delay, severity, geography, ship selection, congestion, and controls. Low traffic may be evidence that the market already considers the risk unacceptable.
For commercial decisions, the goal is not to manufacture an exact probability. It is to prevent a superficially reassuring count from overruling the richer evidence about who sailed, who waited, what happened, and how conditions changed.
This article distinguishes verified source material from ZeaClub analysis, recommendations, forecasts and hypothetical scenarios.
Sources
- IMO, Data on transits through the Strait of Hormuz, March to September 2026Accessed 2026-11-19
- IMO, Middle East information and incident-validation methodologyAccessed 2026-11-19
- JMIC Advisory Update 113, 12 November 2026Accessed 2026-11-19
- JMIC Advisory Update 031, 12 April 2026Accessed 2026-11-19
- UKMTO, Recent incidentsAccessed 2026-11-19
- UNCTAD, Strait of Hormuz disruptions: soaring fuel prices and the burden of inflationAccessed 2026-11-19