How bexy builds its shipping statistics

Where the data comes from

Every shipment figure on bexy comes from Evergreen's public cargo-tracking records. bexy reads a Bill of Lading's record when it first finds it, and once the shipment's estimated arrival has passed it re-checks the record in hourly batches until Evergreen reports the shipment at its final destination, or until 30 days after the estimated arrival if it never does. A shipment someone watches for email alerts is also re-checked about every 12 hours until the watch ends. Searching for a shipment also re-reads it when bexy last read it 6 hours ago or more, within an hourly limit on such re-reads, unless a status already reports the shipment at its final destination or bexy has taken its final reading of it.

Vessel, port and date details are Evergreen's own statements; bexy does not use AIS vessel positions. Route, port, port-to-port route, country, carrier and container owner-code pages are statistics computed from the shipments bexy holds; a shipment page shows one Bill of Lading as bexy last read it. Port codes, positions and time zones on port pages are reference data, not Evergreen's: from the UNECE UN/LOCODE list, Wikidata and the IANA time-zone database (see Definitions).

What the data covers

Only Evergreen (EGLV) shipments are tracked. Other carriers' Bill of Lading prefixes are recognised for reference, but their shipments are not in the data.

bexy finds Bills of Lading by checking Evergreen's Bill of Lading number ranges, and adds any new Evergreen Bill of Lading a visitor looks up. The data is therefore a large sample of Evergreen bookings: not a random sample, not every Evergreen booking, and not a measure of port throughput or market share.

Shipments are counted per Bill of Lading, not per container or TEU. A trade lane runs from the country of the place of receipt to the country of the place of delivery on the Bill of Lading; shipments that start and end in the same country are left out of the lanes. A port page counts a shipment once where it was loaded (a departure) and once where it was discharged (an arrival), so its totals are shipment legs. A port-to-port route counts the shipments loaded at one port and discharged at another, whatever their trade lane. A container owner-code page counts the Bills of Lading carrying at least one container with that code, and their containers: a load is one container on one Bill of Lading, and a box is one container number, which travels under many Bills of Lading over its life.

Definitions

Shipment

One Bill of Lading. A Bill of Lading can cover several containers.

Loaded-on-board date (ETD)

The date in Evergreen's record for loading the cargo on board at the port of loading. Every transit figure starts here, so time spent before loading, such as moving cargo from an inland place of receipt to the port, is not included.

Transit to the final ETA

From the loaded-on-board date to Evergreen's estimated arrival (ETA) at the final place of delivery or, when Evergreen gives no ETA, the last planned arrival in its routing (the Plan Moves list) — the same fallback the shipment page shows. When the place of delivery is inland, this includes the inland leg after the port of discharge. An estimate. Lane, country and carrier estimated transit times, and the outbound figures on port pages, use this definition unless the page says otherwise. A shipment page's scheduled transit is the same span for that one Bill of Lading, to the arrival date the page shows.

Planned port-to-port transit

From the loaded-on-board date to the first planned arrival at the port of discharge in Evergreen's planned routing (the Plan Moves list on the Bill of Lading) — the vessel's call there, not a later row for the same city such as an inland delivery that shares its name — with no inland leg. Used for port-to-port route pages, the port pairs on lane pages and the planned inbound figures on port pages.

Actual transit to discharge

From the loaded-on-board date to the earliest day a container on the Bill of Lading was reported discharged at the port of discharge: a container status that starts with "Discharged" and names that port, read from Evergreen's current container status or the container history bexy records when it re-checks a shipment. A reported event, not an estimate, and it ends at the port of discharge, with no inland leg, so it is not directly comparable with the transit to the final ETA. Route, port-to-port route, country and carrier pages, and the inbound figures on port pages (to the port as port of discharge), show it next to the estimate, labelled as actual, when enough shipments carry it. A shipment page shows its own actual transit when its record holds such a discharge and an on-board date, and the span between them is within the limits under How the statistics are calculated.

Schedule slip vs plan

Per shipment, the day a container was discharged at the port of discharge minus the first planned arrival there (both as defined above), only for shipments that have both; negative when the containers were discharged ahead of plan. Both are whole-day spans from the same loaded-on-board date, so each slip is exact to within a day either way, and a slip of more than 60 days early or late is treated as a data error and dropped. Route, port-to-port route, country, carrier and port pages show its median and middle half, with the number of shipments and sailings, when at least 20 shipments across at least 5 sailings back it, and the port pairs on route pages their median slip on the same floors. Every shipment is measured against its own plan, so it is the cleanest measure of how far schedules slipped.

Sailing

One vessel voyage: the vessel and voyage Evergreen plans for the arrival at the port of discharge, read from the same Plan Moves row as the planned arrival, compared regardless of case and spacing. Every container on a late vessel is late together, so a schedule slip counts the distinct sailings behind it as well as the shipments. A shipment whose vessel is not named yet ("To be advised", or blank) counts toward the shipments of a slip, never toward its sailings.

Transshipment port

A planned stop in Evergreen's routing after the port of loading and before the port of discharge that is not the port of loading, the port of discharge, the place of receipt or the place of delivery. Stops without a name ("To be advised") are ignored. A shipment with no port of loading or discharge, no planned routing, or a routing that never reaches the port of discharge is left out of the transshipment share. This is the plan when bexy read the Bill of Lading, and routings can change after booking. A shipment page labels its own routing as planned direct or names its planned transshipment ports by the same rule, and only when every planned stop between the two ports is named and its voyage route shows the same transshipment calls.

Container owner code (BIC code)

The first four letters of an ISO 6346 container number: three owner letters registered with the Bureau International des Containers (BIC) and the category letter U for a freight container. Owner-code pages cover the codes the BIC register lists for Evergreen Group companies, and name the registered company; boxes from container lessors on Evergreen Bills of Lading carry their own owners' codes. Only well-formed numbers (four letters and seven digits) are counted, and a code's share is its loads out of every well-formed load on the Bills of Lading in question. A box's size and type are Evergreen's own notation, such as 40'(SH): the length in feet, then Evergreen's type code, described only where Evergreen states what the code means.

UN/LOCODE, position and time zone

A port page shows the port's UN/LOCODE, the five-character location code in the code list of the United Nations Economic Commission for Europe (UNECE), release 2024-2, for the city and country in Evergreen's record: the row of the same name, the list's own variant of it (Ad Dammam for Dammam), or a row matched by hand where the list uses another name for the place (Antwerpen for Antwerp). Where the list gives a familiar code to the city's airport (CNSHA, for Shanghai), the page says so. The position is that of the port's own item in Wikidata where bexy found one under the port's UN/LOCODE, else the position the list gives, to the nearest minute, else the position of the city or town in Wikidata, and the page says which; it stands for the port or its city, not a terminal or berth. The time zone is the IANA time zone of that place. Nearby ports are up to 5 other port pages with a known position within 500 km of it, nearest first, by great-circle distance, not sailing distance. A port without a country code, one whose code bexy could not match with confidence, or one published since bexy last matched the list shows none of these.

How the statistics are calculated

Each shipment's transit is rounded to the nearest whole day (a span under half a day counts as one day). Spans of zero days or less, or of more than 120 days, are treated as data errors and dropped.

Medians and percentiles are read from a day-by-day histogram of those spans using the nearest-rank method: the Nth percentile is the smallest number of days that at least N% of the shipments fall within. Pages show the median (50th percentile), the middle half (25th to 75th percentile) and the 90th percentile.

No transit figure is published when fewer than 20 shipments with a usable on-board date and estimated arrival back it, and each figure states how many shipments it rests on. A port pair's planned transit needs 20 shipments with both dates, an actual transit 20 shipments with a recorded discharge at the port of discharge, a schedule slip (of a route, a port pair, a country, a carrier or a port) 20 shipments with both a planned arrival and a recorded discharge, across at least 5 sailings, and a lane's or port pair's transshipment share needs 20 shipments with a routing that can be classified.

Actual transit uses the same rounding, limits and floor as the estimates, but is never mixed with them: a page shows it as its own figure, with its own shipment count. A shipment page's own scheduled and actual transit are that one shipment's spans, rounded and limited the same way; no floor applies to them.

Country and carrier figures pool the trade lanes behind them, and port figures pool the loading ports (inbound) or destination countries (outbound) behind them. Inbound port figures are pooled by port of loading rather than by origin country because loading ports in one country can sit days apart from the same port of discharge, and the planned and the actual figure rest on different shipments: weighting each loading port by its own arrivals gives a loading port that counts in both figures the same weight in each. A loading port can still count in only one of them when it has fewer than 20 measured shipments in the other, so the two published figures can rest on different loading ports, and the planned-versus-actual note is never read off them (see below). Arrivals with no port of loading on record, or one naming the port of discharge itself, count toward the shipments a figure must cover, never toward the figure itself. Because bexy measures a different share of each lane's shipments, a pooled figure weights each lane (or loading port, or country) by its shipment volume instead of simply adding up the measured shipments: only lanes with at least 20 measured shipments of their own count, each lane's measured shipments are scaled up to its full shipment count, and the percentiles are read from that weighted histogram the same nearest-rank way. A pooled figure is published only when the lanes that count hold at least 50% of the pooled shipments. It states the shipments actually measured on those lanes, never the weighted total, and the share of the shipments the lanes cover. Estimated and actual transit are pooled by the same rule.

Where a route, country or carrier page publishes both an estimated and an actual median, it adds a note only when the actual median is at least 15% shorter than the estimate, by at least 3 days but no more than 14 days: the estimate runs to the final place of delivery and can include an inland leg, while the actual stops at the port of discharge. A wider shortfall gets no note, because it is more than an inland leg explains: the two figures can rest on a different mix of routes. An actual median longer than the estimate is never read as slipped schedules, for the same reason: the two figures rest on different shipments.

Whether schedules slipped is judged on the schedule slip instead, on route, port-to-port route, country, carrier and port pages alike: each shipment's discharge day minus its own first planned arrival day at the port of discharge, over only the shipments that have both. The note is shown when at least 20 such shipments across at least 5 sailings back it and their median slip is at least 3 days late, and it states that median and the number of shipments and sailings. The published planned or estimated and actual figures rest on different shipments, and a different mix of routes between them can open or close a gap of several days on its own; measuring every shipment against its own plan removes that. A route, country or carrier page can show the inland-leg note and the slip note together: they measure different things, and containers can be discharged late against the port-side plan even where the actual figure is shorter than an estimate that includes an inland leg. Port and port-to-port route pages compare the actual with the planned port-to-port transit, which has no inland leg, so they note only a slip.

The slip is a plain count, never weighted by volume: a route's slip adds up the slips of all its port pairs, a country's the slips of its routes, a carrier's covers its shipments on international routes, and a port's covers its arrivals whatever the loading port. It is a difference against each shipment's own plan, which does not carry the length of its route the way a transit time does, and shipments with both dates are few, so weighting would scale a handful of them from one route or loading port up to all of its shipments. Slips of more than 60 days early or late are treated as data errors and dropped.

A slip is counted per shipment but happens per sailing: every container on one vessel voyage is discharged late together, so dozens of shipments on one or two late sailings are one or two observations, not a trend. Each slip therefore also counts the distinct sailings behind it — the vessel and voyage Evergreen plans for the arrival at the port of discharge — and is published only when its shipments came on at least 5 sailings, on top of the 20-shipment floor; a page states both counts. A route's sailings are counted across all its port pairs and a country's across all its routes, so a voyage that serves several of them counts once. A shipment whose vessel is not named yet ("To be advised", or blank) still counts toward the shipments, but not toward the sailings.

Some figures are an arithmetic mean (average) rather than a median: the outbound average on port pages, the range of lane averages on country pages, the approximate "~N days" figures in route lists, and any page built from data that predates the percentiles. A mean uses the same shipments and the same 20-shipment floor.

Which pages are published

A lane, port, port pair, country, carrier or container owner code without enough data has no page of its own rather than a thin one.

Trade lanes

At least 25 shipments; at most the 800 busiest lanes.

Ports

At least 50 shipment legs (departures plus arrivals), at most the 400 busiest ports, and real content on at least one side: destination countries backed by at least 25 departures or a published outbound transit, or origin countries backed by at least 25 arrivals or a published inbound transit.

Port-to-port routes

At least 25 shipments loaded at one port and discharged at another, both ports with pages of their own and in different countries; a published planned port-to-port or actual transit (at least 20 shipments); and a shipment loaded on board within the last 365 days, so a service that stopped sailing drops out; at most the 400 busiest pairs.

Container owner codes

A code the BIC register lists for an Evergreen Group company, on at least 100 tracked Bills of Lading, one of them loaded on board within the last 365 days, with at least one lane or port page to link to.

Country pages

At least 2 published trade lanes from the country (for a "shipping from" page) or to it (for a "shipping to" page).

Carriers

At least 50 Bills of Lading with the carrier's prefix and at least one known lane or loading port; at most 30 carriers. Only Evergreen is tracked, so Evergreen is the only carrier with a page.

How often the statistics update

The statistics are rebuilt once a night from a single snapshot: every dataset counts the same shipments, the ones bexy held when the build started. The lane, port, port-to-port route, container owner-code and carrier data are replaced together or not at all. A build that comes back empty or loses more than 50% of a dataset's pages is refused, and the previous night's figures stay up.

Every statistics page shows the date of the data it was built from; the current data is dated 2026-10-19.

Limitations

Estimates, not arrivals

Apart from actual transit, transit times are worked out from Evergreen's dates, which are estimates, not measured arrivals.

Actual transit covers few shipments

Evergreen's record shows a container's discharge only until the container moves on, and for most Bills of Lading bexy first reads the record after the container has been returned empty. Actual transit therefore rests on far fewer shipments than the estimates, and on the shipments bexy happened to read between discharge and pick-up, which need not be typical; in the current data, 71 of the 252 published trade lanes publish an actual transit, resting on 19,033 shipments with a recorded discharge.

ETAs change

Evergreen revises ETAs, and bexy re-reads a shipment only after its ETA (or, without one, its last planned arrival) has passed — unless someone watches it for email alerts, which re-reads it about every 12 hours, or someone searches for it, which re-reads it, within an hourly limit, when bexy last read it 6 hours ago or more, unless a status already reports it at its final destination or bexy has taken its final reading of it — so a figure can rest on an ETA that was later revised.

Counts are what bexy has found

Counts reflect the Bills of Lading bexy has found, not Evergreen's total volume or any carrier's market share. A higher count can mean bexy found more Bills of Lading, not that trade grew.

Routings are plans

Planned routings and transshipment ports are Evergreen's plan when bexy read the Bill of Lading, not how the cargo actually moved.

Ports by city name

Ports are identified by the city name in Evergreen's record, so two ports with the same name in different countries can be counted together. The UN/LOCODE, position and time zone a port page shows are those of the port in the country the page names.

How to cite

Every route, port-to-port route, port, country, carrier and container owner-code page has a "Cite this page" line with the page title, the date of its data, the number of shipments it rests on and its URL. Port pages count shipment legs and carrier pages Bills of Lading, and their lines say so. Figures change nightly, so cite the page you took a figure from, with its date.

Frequently asked questions

Where does bexy's shipping data come from?

From Evergreen's public cargo-tracking records. bexy reads each Bill of Lading's record when it first finds it, and once the shipment's estimated arrival has passed it re-checks the record in hourly batches until Evergreen reports the shipment at its final destination, or until 30 days after the estimated arrival if it never does. Vessel, port and date details are Evergreen's own statements, not AIS vessel positions. Port codes, positions and time zones on port pages are reference data from the UNECE UN/LOCODE list, Wikidata and the IANA time-zone database.

Does bexy cover every carrier and every shipment?

No. Only Evergreen (EGLV) shipments are tracked. bexy finds Bills of Lading by checking Evergreen's Bill of Lading number ranges, so the data is a large sample of Evergreen bookings: not a random sample, not every Evergreen booking, and not a measure of port throughput or market share.

How is transit time measured?

Unless a page says otherwise, transit is estimated and runs from the loaded-on-board date to the carrier's ETA at the final place of delivery, or the last planned arrival when the carrier gives none, which can include an inland leg after the port of discharge. Planned port-to-port transit ends instead at the first planned arrival at the port of discharge in Evergreen's routing. Both are the carrier's estimates, not measured arrivals. Actual transit, shown separately on route, port-to-port route, country and carrier pages and inbound on port pages, labelled as actual, runs from the loaded-on-board date to the day a container was discharged at the port of discharge, as reported in the carrier's container events. A shipment page gives the same spans for that one Bill of Lading: its scheduled transit to its own arrival date and, when its record holds a container status naming a discharge at the port of discharge, its actual transit, both within the same limits.

Why is actual transit shown on so few pages?

Evergreen's record shows a container's discharge only until the container moves on, and for most Bills of Lading bexy first reads the record after the container has been returned empty, so few shipment pages can show their own actual transit. A statistics page therefore publishes actual transit only where at least 20 shipments carry a recorded discharge at the port of discharge, and it rests on far fewer shipments than the estimates; in the current data, 71 of the 252 published trade lanes publish an actual transit, resting on 19,033 shipments with a recorded discharge.

Why do some pages show no transit time?

A statistics page publishes a transit figure only when at least 20 shipments with a usable on-board date and estimated arrival (or, for actual transit, a recorded discharge) back it, and a shipment page shows its own only when its record has both dates. Spans of zero days or less, or of more than 120 days, are treated as data errors and dropped before counting.

How are the median and percentiles calculated?

Each shipment's transit is rounded to the nearest whole day, and the percentiles are read from that day-by-day histogram by the nearest-rank method: the median is the smallest number of days that at least half of the shipments fall within, and the 90th percentile the smallest that covers 90% of them. Country, carrier and port figures that pool several trade lanes (or, on port pages, several loading ports inbound or destination countries outbound) weight each one by its shipment volume, so a lane where bexy happened to measure more shipments does not outweigh a busier one: only lanes with at least 20 measured shipments count, each lane's measured shipments are scaled up to its full shipment count, and the figure is published only when the lanes that count hold at least 50% of the shipments. The shipment count it states is the shipments actually measured.

Why can actual transit differ from the estimated transit?

The estimate is the carrier's plan as bexy read each Bill of Lading and runs to the final place of delivery; actual transit runs to the day a container was discharged at the port of discharge. The two also rest on different shipments, since only some carry a recorded discharge, so the gap between them is never read as slip. Where both medians are published and the actual median is at least 15% shorter than the estimate, by at least 3 days but no more than 14 days, the page notes that the estimate can include an inland leg after the port of discharge, which the actual leaves out. A wider shortfall gets no note: an inland leg alone rarely explains it, and the two figures can rest on a different mix of routes. Whether schedules slipped is judged on the schedule slip instead: each shipment's recorded discharge is compared with its own first planned arrival at the port of discharge, over only the shipments that have both. Route, port-to-port route, country, carrier and port pages note a slip when at least 20 such shipments across at least 5 sailings back it and their median slip is at least 3 days late, and the note states that median and the number of shipments and sailings. A route, country or carrier page can show both notes: they measure different things, and containers can be discharged late against the port-side plan even where the actual figure is shorter than an estimate that includes an inland leg.

What is the schedule slip vs plan?

The days between the carrier's first planned arrival at the port of discharge and the day a container was actually discharged there, worked out per shipment and only for shipments that have both: discharge day minus planned arrival day, so a negative slip means the containers were discharged ahead of plan. Both are whole-day spans from the same loaded-on-board date, so each slip is exact to within a day either way, and a slip of more than 60 days early or late is treated as a data error and dropped. Route, port-to-port route, country, carrier and port pages publish its median and middle half, with the number of shipments and sailings, when at least 20 shipments across at least 5 sailings back it, and the port pairs on route pages their median slip on the same floors. A sailing is one vessel voyage, read from the vessel and voyage Evergreen plans for the arrival at the port of discharge. Every container on a late vessel is late together, so the shipments on one sailing are one observation, not many: without the sailing floor, one or two delayed ships could read as a trend. A shipment whose vessel is not named yet ("To be advised") counts toward the shipments, never toward the sailings. It is the cleanest measure of slip because every shipment is measured against its own plan: the published planned or estimated and actual figures rest on different shipments, and a different mix of routes between them can open or close a gap of several days on its own. For the same reason it is a plain count, never weighted by volume: a route's slip adds up its port pairs', a country's its routes', a carrier's covers its shipments on international routes and a port's its arrivals, whatever the loading port. A slip does not carry the length of its route the way a transit time does, and weighting would scale a few paired shipments up to a whole route's or loading port's volume.

How often are the statistics updated?

Once a night, from a single snapshot of the shipments bexy holds; the lane, port, port-to-port route and carrier data are replaced together or not at all. Every statistics page shows the date of its data; the current data is dated 2026-10-19.

How do I cite bexy?

Use the "Cite this page" line on the route, port-to-port route, port, country or carrier page the figure comes from. It gives the page title, the date of its data, the number of shipments it rests on and the page URL.