How every number is derived
The dashboard and the workbooks it exports show figures computed from three kinds of base data. This page states, for each number, the rule that produces it — the population, the statistic and every threshold — so an exported figure can be traced back to the records behind it.
3Base record sets
10Records before a median
DailyPipeline rebuild
~2 moContainer record lag
The base data
Three record sets. Every figure on the site names which one it was measured on.
AIS sailingsVessel movements from AIS positions, paired into departure → arrival sailings between two ports, with the track along the way, the ship, its capacity, the operator and the service it was deployed on. Exported as the AIS sailings sheet.
AIS port callsEach arrival and departure of a vessel at a port or terminal, with its stay, draft and the service the vessel was on at the time.
Container Tracking recordsCarrier event records per container, not per ship: gate-in, loading, discharge, transshipment and gate-out, at the port and terminal they happened. AIS cannot see a box, which is why these are a separate record set.
Figures are rebuilt from these records by a nightly pipeline. Every card that depends on the record’s edge says so — Last call, Data as of — and every figure on that page counts back from that date.
The two layers do not share a horizon. Recorded sailings and port calls arrive daily, so vessel figures are current to yesterday. Container Tracking records trail by roughly two months: a container’s events are only complete once its journey ends, so the most recent weeks are always still filling in. That is why a dwell figure is a median over years of records with its count shown, never a reading for this week — and why the two are never added together or compared as if they covered the same period.
Every figure, and the rule behind it
One row per number shown on the dashboard: what it counts, how it is computed, and which record it comes from.
Route lookup
The figures beside the map
| Number | How it is derived | Base data |
|---|---|---|
| Recorded voyages | The count of recorded sailings from the origin to the destination. With nothing selected on a direct lane it is every sailing in the lane’s index; with a service, a leg or a time window selected it is the sailings drawn for that selection; on a relay it is the sailings on the scarcer of the two legs. Nothing sails origin → hub → destination as one voyage, so there is no end-to-end journey count to give; the chain can only run as often as its thinner leg, and that is what the card bounds. The full evidence is per leg, and the legs carry it. | Recorded sailings |
| Vessels | Distinct ships that sailed the lane in the last 18 months, from the same index the voyages come from. Under a service selection, the distinct ships on that service’s sailings. | Recorded sailings |
| Ports called | Intermediate ports called on at least 25% of the lane’s sailings (“common stops”); the detail line names the top three with their share. On a relay: the typical number of stops per leg. | Recorded sailings |
| Distance | Nautical miles actually sailed along the recorded tracks — not a great-circle line. Lane-level it is the median across sailings; with a selection it is the mean per leg over the drawn tracks, legs summed. | Recorded vessel positions |
| Median transit | The median port-to-port days (departure to arrival) of the sailings in view. On a relay with nothing selected it is the route-finder’s typical: leg medians + hub wait + hub bias. | Recorded sailings |
| Speed made good | Distance ÷ (sea days × 24), in knots. Sea days only — time spent in a hub is never in the denominator. | Recorded sailings |
| Timeline | The sailings in view, bucketed by departure date into equal time bins across the record. Brushing a window narrows every figure on the page to it. | Recorded sailings |
| Map | The recorded positions of the sailings in view. On a busy lane with nothing selected only the top services are drawn — most sailings first, until 60% of the lane’s sailings are covered, at most five, each with at least 10 sailings — and the caption says so. | Recorded vessel positions |
| Coverage strip | “Ports” is the number of ports in the lookup index; “Lanes drawn” is the number of lanes in the corridor catalogue. | Recorded sailings |
The gate-to-gate strip
| Number | How it is derived | Base data |
|---|---|---|
| Gate in | The median number of days a container waited at the origin terminal between gate-in and loading, over every carrier’s gate records for that port (the record count is shown). Lane-wide — gate records carry no service, so this is never split by service. | Container Tracking records |
| At sea | The median sailing days of the slice in view: the whole lane, or the leg, or the selected service on that leg. | Recorded sailings |
| Transship dwell | The median days between discharge and reload at the hub, over every carrier’s transshipment records at that hub; the P25–P75 band and record count are shown. | Container Tracking records |
| Gate out | The median days between discharge and pickup at the destination terminal, over every carrier’s gate records for that port. | Container Tracking records |
| Gate to gate | Gate in + sea legs + hub wait + gate out; on relays, plus the hub bias (below). It is shown only when every stage in the frame measures the same slice — a service that serves one leg of a relay is not summed with lane-wide figures for the rest. | All three |
| Proportion bar | Each stage’s share of the gate-to-gate total, in days. The striped segment is the hub bias. | — |
The routings list
| Number | How it is derived | Base data |
|---|---|---|
| Direct / via | A “via” routing is a change of vessel at that hub, not an intermediate call. A ship that calls five ports on the way is still direct: the box never moves. | Recorded sailings |
| Typical (d) | Median leg days (both legs on a relay) + median hub dwell + hub bias. | Recorded sailings + Container Tracking records |
| Fast / Slow | Fast = P25 of each leg + P25 dwell, with no bias. Slow = P75 of each leg + P75 dwell + bias. Spread = slow − fast. The band is deliberately asymmetric and is not a confidence interval. | Recorded sailings + Container Tracking records |
| Hub bias | The measured shortfall of leg + dwell + leg against real end-to-end journeys through that hub — a median correction, typically +4 to +15 days. A hub with no measured bias receives the median across measured hubs (6.3 d) and its rows are marked “hub wait estimated”. A hub with no measured dwell at all is not scored on total time. | Container Tracking records |
| /mo and voy | Sailings per month, and sailings recorded, on whichever leg has fewer. | Recorded sailings |
| Ranking | Routings are ranked by evidence tier first, then by typical days. A routing competes for the top only if its hub wait was measured on this lane, it has at least 10 recorded voyages and at least 2 sailings a month; the rest are listed below the divider with the reason. The badge reads “Fastest”, or “Best evidenced” when a thinner row is quicker on paper. | — |
Lines
| Number | How it is derived | Base data |
|---|---|---|
| Carrier cards | MEDIAN is that carrier’s median transit on the leg; VOYAGES its sailings on the leg. | Recorded sailings |
| Services | Each service under a carrier lists its sailings, median and P25–P75 band. The label is Strong at 25 or more sailings, Moderate at 10 or more, Thin below 10 (folded away); a service needs 3 sailings to be listed at all. | Recorded sailings |
Service lookup
| Number | How it is derived | Base data |
|---|---|---|
| Attribution | Every recorded port call is attributed to the service the vessel was deployed on at the time of the call. This is a record of what actually sailed, not a published rotation, and service tags mature with a lag. | Recorded port calls |
| Rotation ports | Ports called on at least 33% of the service’s loops in the last 12 months, with a minimum of 3 calls. A loop is one call at the service’s busiest port. Ports called on fewer loops are listed as “below threshold”. | Recorded port calls |
| Vessels | Distinct ships with calls attributed to the service. | Recorded port calls |
| Calls recorded | Every attributed call since the service was first seen. | Recorded port calls |
| Median vessel size | The median nominal TEU across the service’s recorded calls, not across its distinct ships: a vessel that calls thirty times weighs thirty times. A median over the distinct hulls is a different figure and reads higher. | Vessel registry |
| Round trip | The sum of the median days between consecutive rotation ports plus the median port stays. Prefixed “≥” when a hop is unmeasured: a lower bound. | Recorded port calls |
| Last call | The newest attributed call; every figure on the page counts back from this date. A service is Active when its newest call is within 365 days of the record’s own edge — measured against the record, not today’s date, because a service is not dead because our data stops. A string whose last call falls outside that is marked ended rather than listed as if you could still book it. | Recorded port calls |
| Port calls chart | Calls per port inside the brushed window; qualifying rotation ports are ringed on the map. | Recorded port calls |
| Vessels table | TEU, length overall, last port and when each ship was last seen on the service. Ships last seen more than 180 days ago are greyed. | Recorded port calls + vessel registry |
Port lookup
| Number | How it is derived | Base data |
|---|---|---|
| Calls 12m | Commercial container-vessel calls with an arrival in the last 12 months. | Recorded port calls |
| TEU 12m | The nominal capacity of those calls, summed. | Recorded port calls + vessel registry |
| Unique vessels | Distinct ships among those calls. | Recorded port calls |
| Gate-in / Gate-out / Transship dwell | Medians over the container records at this port, with the P25–P75 band and record count. Not quoted below 10 records. | Container Tracking records |
| Latest calls | The most recent recorded calls: vessel, terminal, arrived, departed. | Recorded port calls |
| Connections | Ports linked to this one by direct sailings in the last 12 months; the top five by sailings are shown, with the total. | Recorded sailings |
| Cargo mix | The share of recorded containers transshipped at the port versus gateway cargo (everything that is not transshipped). | Container Tracking records |
Conventions that apply everywhere
Five rules that hold on every page, every card and every exported sheet.
- Medians, not means. A typical figure is the median of its population; the one exception is the per-leg distance under a selection, which is a mean of the drawn tracks.
- P25–P75 is the middle half of the records — the band a quarter fall below and a quarter above. It is not a confidence interval.
- Thin samples are shown, not ranked. Below 10 records a figure is listed with its count but never quoted as a median, never badged and never sorted above a well-evidenced one.
- One grain per figure. Sailings, port calls and containers are different populations. No card adds or compares them, and an exported sheet names which one it counts.
- Selections narrow, never recompute. Picking a service, a leg or a time window filters the same records; the rule behind each figure does not change.
The exports
The workbook rebuilds exactly what the page shows: one sheet per pane, in the state the page was in at export time, with the summary figures first. Its first sheet, How derived, repeats the rules above for that subject. Aggregate sheets are never truncated. Downloads are counted as files, not rows — one workbook is one download however large it is. Free includes one a month, Pro five, Enterprise unlimited, resetting on the 1st.
See the numbers in place
Open any lane, service or port on the dashboard — every card there is one of the figures described above.
Every export carries these rules on its own “How derived” sheet — [email protected]