For Agents

Reference for the Phase-2 play only: a real-time AIS feed to nowcast PortWatch next published number before its weekly Tuesday release. Do NOT use any of these as the settlement source (that is basis risk — see hormuz-bet-basis-risk). The provider landscape has consolidated into Kpler (now owns MarineTraffic, Spire, FleetMon — enterprise quote-only). Cheap self-serve REST options exist but are all terrestrial-only (doubly exposed mid-strait) and emit raw vessel positions you must aggregate into strait crossings yourself.

The market landscape for ship-tracking data, scoped to its only legitimate use here: predicting the settlement number ahead of publication.

Scope: Phase-2 Only

These feeds are a leading indicator, never the settlement source

The market resolves on PortWatch (see hormuz-bet-resolution-mechanics). A third-party AIS feed is useful ONLY to predict how PortWatch will count this week before it publishes — you must learn the mapping from the feed to PortWatch number, not assume they are equal. Using one as the resolution source is the basis-risk trap documented in hormuz-bet-basis-risk.

The Enterprise Consolidation: Kpler

The independent-provider market has collapsed into Kpler, which now owns MarineTraffic, Spire, and FleetMon. These are enterprise, quote-only (no cheap self-serve), so they are out of scope for a small project unless a paid plan is justified later.

Cheap Self-Serve Options

ProviderPriceHistoryAccessNotes
Datalastic~EUR 99/mo~2 yrREST, inradius bboxNo native line-crossing; you build crossings from positions.
MyShipTrackinglow~2 yr, 3000 rec capRESTTerrestrial-only.
aisstream.iofreenone (live only)WebSocketFree live stream, no history.
Global Fishing Watchfree~2012-presentEvents APIIncludes AIS-gap / dark-event detection — useful for modelling evasion.

All the cheap ones are terrestrial-only AIS

Terrestrial AIS receivers have limited offshore range, so mid-strait coverage is weak — exactly where Hormuz transits happen. Combined with the dark-vessel/jamming problem (see hormuz-bet-basis-risk), cheap feeds are doubly exposed here. They also emit raw vessel positions, so you must aggregate them into strait crossings yourself (define a gate line across the strait, detect transits, dedupe).

Implication for Phase 2

If Phase 2 is pursued, the realistic cheap-feed build is: ingest live positions (Datalastic inradius around the strait, and/or aisstream live), define a crossing gate line, count transits, and learn a mapping from that home-grown count to PortWatch eventual published number. Global Fishing Watch dark-event detection can be a covariate to correct for evasion. None of this is needed for Phase 1, which only models the free PortWatch history (see hormuz-bet-portwatch-data, hormuz-bet-backtesting).