Predictive ETA vs the carrier's ETA

A predictive ETA is an arrival estimate produced by a model rather than by the carrier. It is built from where the vessel actually is, how fast she is going, and how long the berth queue at the discharge port has been running. It is a second opinion, not a replacement for the line’s own number.

It exists because the carrier ETA has a known failure mode: it is often the published schedule until somebody updates it, and nobody updates it the moment a vessel starts waiting at anchor.

What it is calculated from #

Vessel movement from AIS, the rotation the ship still has to complete, and congestion at the destination — how many vessels are queuing and how quickly the terminal has been working them. Some models add weather and the line’s own historical reliability on that service.

Which is also the list of its weaknesses. A model reading an interpolated position produces an estimate built on an estimate, and a model that has never seen the disruption now happening extrapolates from a world that has changed.

Where it helps, and where it does not #

It is most useful in the last stretch, when the question has stopped being “where is the ship” and become “when will the terminal work her”. That is a queue, and a queue is exactly what a model can see and a published schedule cannot.

It is least useful after a transshipment. The honest answer there is not a better number: your container may be on a different vessel entirely, and no estimate for the original ship is about your fruit.

And it changes nothing about your clocks. Free time starts at discharge, which is an event, not a forecast. A prediction is for deciding whether to move a haulier booking today; the event is what you commit against.

Trackberry shows the provider’s prediction underneath the carrier’s ETA rather than instead of it, and only in the window where it is worth reading. The tracking card carries both, so what you see is the comparison.

FAQ #

Should I plan against the predictive ETA or the carrier’s? #

Plan provisionally against either, commit against the discharge event. Where the two disagree by more than a day close to arrival, the gap itself is the signal — it usually means congestion the carrier’s schedule has not absorbed yet.

Why does the prediction differ from the line’s own date? #

Because they are built from different things. The line’s is a schedule adjusted by its operations; the prediction is derived from observed vessel behaviour and the queue at the destination. Late in a voyage, the second one often knows about a delay first.

Is a predictive ETA more reliable? #

Sometimes, and not by default. Its quality depends on the position data underneath it and on how ordinary the current conditions are. Read it as a second opinion, and watch the movement of both numbers rather than trusting either value alone.