Methodology

The headline output is the Strait State: OPEN, DISRUPTED, or CLOSED. Alongside the state we publish a Tension Level (NORMAL / ELEVATED / CRITICAL) and a Threat Score (0–100) that fuses timeline event severity with market volatility. An analyst confidence value in [0, 1] grows with source diversity and event density.

The map shows live vessel positions from AISStream.io when available, plus representative vessel dots animated along the IMO shipping lanes — proportional to the day's IMF PortWatch transit count by vessel type (tanker · cargo · container · dry bulk).

Status is derived in app/lib/api.ts → deriveStatus(). Runs client-side on every data refresh using the latest timeline events and Brent price change.

Step 1 — Keyword state detection

# Scan events from the past 72 hours
CLOSURE_PATTERNS = /closed|closure|shut down|blocked|blockade|suspended|halt/
PARTIAL_PATTERNS = /diverted|rerouted|delayed|evacuated|traffic disrupt/

if any recent event matches CLOSURE_PATTERNS → state = CLOSED
elif any recent event matches PARTIAL_PATTERNS → state = DISRUPTED
else → state = OPEN

Step 2 — Multi-signal Threat Score (0–100)

# Timeline Severity Score (last 24 h)
weight = { low: 1, medium: 2, high: 4, critical: 7 }
raw_timeline = sum(weight[e.severity] for e in last24)
timeline_score = min(100, raw_timeline × (100 / 35))   # normalised 0–100

# Market Volatility Score
# Brent spike > 2% starts triggering; 5% spike → score 100
if brent_change_pct > 2:
    market_score = min(100, (brent_change_pct - 2) × 33.3)
else:
    market_score = 0

# Final 50/50 blend
threat_score = round(timeline_score × 0.5 + market_score × 0.5)

# Tension level thresholds
if threat_score ≥ 80 or state == CLOSED   → CRITICAL
elif threat_score ≥ 40 or state != OPEN   → ELEVATED
else                                        → NORMAL

# State override — timeline events required to escalate
# (pure Brent spike without events never closes the strait)
if state == OPEN and last24 not empty and threat_score > 85:
    state = DISRUPTED

Step 3 — Confidence

sources    = unique source names contributing to last24 events
confidence = min(0.99,
  0.55
  + 0.06 × |sources|
  + 0.02 × min(|last24|, 6)
)

The hero map renders two independent vessel layers:

• →AIS live dots — real MMSI positions streamed from AISStream.io, refreshed every 15 s. Colour-coded by type. Shown only when the collector is active.
• →PortWatch representative vessels — animated dots distributed along IMO TSS lanes, proportional to the latest daily transit count by type: tanker (amber), cargo (sky), container (purple), dry bulk (slate). These are illustrative — not real GPS positions.

IMF PortWatch updates weekly (~Tuesdays 09:00 ET, 2–3 day lag). Chokepoint ID: chokepoint6. Historical baseline: ~34 transits/day.

• —Timeline / RSS aggregator60 s(KV cached)
• —News (GDELT)5 min(KV cached)
• —Brent · WTI · Henry Hub5 min(KV cached, EIA primary)
• —IMF PortWatch transit counts6 h(KV cached)
• —Weather (Open-Meteo)15 min(KV cached)
• —AIS vessel positions15 s(live WebSocket, KV fallback)
• —Feed health30 s

All routes run on Cloudflare Pages Functions (Edge Runtime) with Cloudflare KV caching.

A read-only JSON API is available for embeds and partner integrations. CORS is allow-all; responses are cached at the edge.

GET /v1/status              # strait state, tension level, confidence, reason
GET /v1/events?limit=30     # latest aggregated timeline events
GET /v1/events?since=ISO    # incremental fetch since ISO timestamp
GET /v1/metrics             # markets (Brent/WTI/NG), weather, event counts
GET /feed.xml               # RSS 2.0 for journalists and aggregators

License: CC-BY-4.0. Attribution: “Global Chokepoints Alerts” with a link to global-chokepoints.pages.dev. Full interactive reference at /docs.

• Keyword matching
  Closure/disruption detection is regex-based. Ironic, quoted, or context-heavy headlines can produce false positives. Treat as directional signal only.
• AIS coverage gaps
  AISStream.io free tier covers only vessels transmitting within range of terrestrial receivers. Military vessels, tankers in radio-quiet mode, and vessels in Iranian waters may be invisible.
• PortWatch lag
  IMF data is 2–3 days lagged and updates weekly. Map vessels reflect the latest complete day, not today. They are illustrative, not real-time positions.
• Yahoo Finance rate limits
  Unofficial endpoint; may return 429 under load. EIA is the primary crude source; FRED for Henry Hub. Yahoo is fallback only.
• RSS deduplication
  URL-based. Near-duplicate stories from different outlets may appear separately.

• v0.6.0 · 2026-05-27
  Full rebrand to Global Chokepoints Alerts — multi-chokepoint intelligence platform (Hormuz, Red Sea, Suez, Panama, Taiwan Strait). Added TickerBar, ChokepointsPanel card grid, HeroStatus CTAs, region inference in Timeline. System-wide flat UI (no rounded corners). Feed health row, relative timestamps in NewsFeed, vessel count in header.
• v0.5.0 · 2026-05-14
  Migrated to Cloudflare Pages (global-chokepoints.pages.dev). Added PortWatch animated vessels on map. KV caching on /api/timeline and /api/weather. Fixed confirmation email token bug. Tightened threat-score false-positive rule.
• v0.4.0 · 2026-05-13
  Replaced Three.js 3D scene with live Leaflet map (CartoDB dark tiles, shipping lanes). Header z-index fix (z-1100 over Leaflet). News feed skeleton loading. Decoupled GDELT fetch from dashboard Promise.all.
• v0.3.0 · 2026-05-12
  IMF PortWatch integration — real daily transit counts. FRED API for Henry Hub (permanent fix for Yahoo 429s). KV-persisted Yahoo cache. AIS binary Blob frame decoder.
• v0.2.0 · 2026-05-11
  Live data architecture: API routes, RSS aggregator, multi-signal status derivation, public /v1 API, RSS feed, methodology page, Cloudflare KV + D1 bindings.
• v0.1.0 · 2026-05-10
  Initial Next.js dashboard; visual design + i18n (EN/PT-BR) only.

This site provides information for situational awareness only. It is not navigational, financial, military, or operational advice. Always verify with official authorities (IMO, NAVCENT, UKMTO, port authorities, your insurer's war-risk cell) before making decisions. Threat scores are probabilistic outputs of an automated algorithm and do not represent a forecast of imminent events.