Bespoke Maritime · Freight · Commodities Intelligence

Custom platforms, research, and intelligence for maritime trade.

DMI builds what you actually need: an operations platform, a scoped study, a recurring brief. Built around your operation, and proven on our own record — we forecast arrivals days to weeks ahead, and publish how wrong we were.

Get an accuracy card

An accuracy card is a one-page graded error record for the ports or vessels you name. Send yours and we send it back free.

Email Luke

A short reply, usually same day.

97
Ports we forecast
349 K+
Forecasts graded
1.5 d
Typical arrival error, 10 days out
2.7 d
Typical arrival error, 4 weeks, bulk

The problem

Every vessel-tracking subscription shows the same screen. The question that costs you money is not on it.

Where the money leaks

Self-reported
The AIS arrival field — the ETA a ship’s own crew types into its transponder — unvalidated and often left blank
$15K–$25K
What one day of a ship waiting at the berth costs, under a typical contract
Days early
When the charter and hedge decisions get made — on whatever number you have
Bulk-terminal gantry cranes over a loading berth at a Gulf-Coast export port
Ships wait offshore for a free berth. Every waiting day is billed to somebody.

What we do

Three ways
we work.

For the desks that plan around ship arrivals: traders and charterers, terminals and agencies, and the operations in between.

  1. 01 Bespoke platforms & analytics
  2. 02 Commissioned research & studies
  3. 03 Standing intelligence
  1. BUILT TO SPEC

    Bespoke platforms & analytics

    Operations platforms and analytical tools built to your spec on the vessel-and-port data DMI already runs. For example: a berth-queue view for one terminal — which ships are waiting, in what order, and when each one is likely to go alongside — built around your own contracts, and a screen no off-the-shelf product would ever include.

  2. QUESTION IN · ANSWER OUT

    Commissioned research & studies

    Scoped studies of ports, flows, routes, and markets. A question goes in. A defensible, written answer comes out.

  3. ON YOUR DESK · RECURRING

    Standing intelligence

    A brief that lands on a set schedule, scoped to your operation: the cargo routes you follow, the ports you ship through, with graded arrival forecasts for the vessels inside it.

A laden container vessel bow-on under way — the per-vessel hull Polaris forecasts to the berth.
One hull under way. Every forecast is made for a single ship like this one, and graded on the hour it ties up.

Ten days out

Anyone can track a ship. We call the arrival ten days before it happens.

Polaris is our arrival‑forecast engine. Ten days out it typically lands within about a day and a half of the real arrival, and even four weeks out a bulk carrier’s forecast typically lands within about 2.7 days, less than a tenth of the time still to run. The AIS field gives you one number and no idea how wrong it is. Ours comes graded against 349,000+ realized arrivals, and that record is the proof behind every piece of DMI work.

Same vessel, ten days out: one guess vs a graded window

A conceptual plot. Time runs left to right, from ten days before arrival to the arrival itself; height is the vessel’s remaining distance to the berth. The Polaris forecast is a shaded gold window that starts wide ten days out and narrows onto the realized arrival. The AIS estimate is a single dashed rule standing apart from that window, with no band around it. A solid rule marks the realized arrival — the hour the ship was actually tied up.

AIS ETA · one guess

one number, no stated range

Realized arrival

the hour the ship tied up

Polaris · calibrated window

narrows as the ship closes

Days before arrival 10 → 0

Distance to berth

Drawn to show the shape of the comparison — the measured figures live in the ledger.

Typical miss at ten days out

Shorter is better.

  1. Polaris

    1.5d

    A graded window with a stated range — and it narrows as the ship closes.

  2. AIS ETA (CREW-ENTERED)

    Not published

    One number, with no error figure to size it against.

Both bars run on the same scale — zero to ten days of arrival error. The hatched bar runs off the end because no figure is published to draw it to.

Typical error at ten days out, as each method publishes it
AIS ETA (CREW-ENTERED)
No error figure published
Polaris
1.5d typical, with a stated range

Only one of the two publishes an error figure at all — which is the comparison.

Accuracy record

Every forecast is scored against what actually happened.

Every forecast is checked against the real arrival — the hour the ship is actually tied up at the berth — and only on voyages the forecast never saw while it was being built.

Each ETA ships as a window, and we publish how often that window is right, so you know how far to trust it before you lean on it.

The figures below are measured across the whole production footprint: eight Gulf Coast ports worked daily, 97 ports in production across the wider trade.

Most accuracy claims in this market arrive with no graded record behind them. This one does.

How we grade

Typical
The middle of the graded errors: half the forecasts miss by less than the figure shown, half by more.
Graded
Scored after the fact against the hour the ship was actually tied up at the berth, on voyages the forecast never saw.

Typical arrival error

  1. 1.5 d Ten days out
  2. 1.9 d Two weeks out
  3. 3.1 d Four weeks out
  4. 2.7 d Of which, bulk carriers only

Every figure above is measured against what actually happened, voyage by voyage, and logged.

Record last graded July 2026

About

A maritime intelligence firm working the Gulf Coast first.

We work one stretch of coast closely enough to be specific about it, and we stay answerable for every number we publish.

DMI is run by Luke Driscoll. When you commission work, he is who you deal with, start to finish. The method is on the table, the Polaris record is graded in the open, and the same record is downloadable above.


Method
AIS tracks run through our own forecasting models, built and tuned in-house. Nothing is resold: the forecasts are ours, and so is the error record beside them.
Coverage
Eight U.S. Gulf Coast ports worked daily — Houston, New Orleans, Corpus Christi and the export gateways along the coast — inside 97 ports running in production across the wider trade.
Who you deal with
Founder-led and independent. You reach the principal who does the work, usually with a same-day reply.
A bulk carrier on the Gulf horizon at dusk, last light on flat water

Start a conversation

Tell us what you are trying to know.

Email Luke

A short reply, usually same day.

Or write to luke@driscollmaritime.com