Index Hierarchy
The platform computes indices at three levels of abstraction, from observable signals (L0) through intermediate constructs (L1) to composite indices (L2).
L2 Composite Indices (3) ├── GRI: Geopolitical Risk Index ├── NII: Narrative Interference Index └── CI: Consistency Index L1 Sub-Indices (6) ├── DVS: Domestic Volatility Score ├── EES: External Escalation Score ├── EFS: Economic Fragility Score ├── AFI: Alliance Friction Index ├── LSI: Legitimacy Stress Index └── IIS: Information Interference Score L0 Observable Indicators (37) ├── DVS indicators (8): protest_intensity, political_violence, ... ├── EES indicators (7): mil_mobilization, border_incidents, ... ├── EFS indicators (7): fx_volatility, cds_spread_delta, ... ├── AFI indicators (6): alliance_statement_sentiment, ... ├── LSI indicators (4): approval_rating_delta, ... └── IIS indicators (5): cross_platform_velocity, ...
GRI: Geopolitical Risk Index
Measures the likelihood of disruptive change affecting diplomatic and economic predictability.
Formula
GRI = 0.30·DVS + 0.25·EES + 0.25·EFS + 0.10·AFI + 0.10·LSINII: Narrative Interference Index
Detects coordination, laundering, and amplification in the information environment.
Formula
NII = 100 · sigmoid(0.45·C + 0.30·L + 0.25·A) · (0.6 + 0.4·V) · (1 + 0.5·S)CI: Consistency Index
Quantifies divergence between stated policy norms and observable actions. Higher scores indicate greater alignment between words and deeds.
Formula
CI = 100 · (1 - D(stated_norms, actions | comparable_cases))Where D is a divergence measure computed against comparable historical cases, accounting for regime type, coalition structure, and international commitments.
Data Pipeline
- 1Evidence Collection
OSINT sources ingested daily: diplomatic cables, economic data feeds, military movement reports, social media streams.
- 2Event & Narrative Extraction
Structured events extracted with severity scoring. Narrative clusters identified with coordination and attribution analysis.
- 3Indicator Computation
37 L0 indicators computed from evidence, normalized to 0-100 scale using z-score transformations against historical baselines.
- 4Index Calculation
L0 indicators aggregated to L1 sub-indices, then to L2 composite indices using weighted formulas with momentum smoothing.
- 5Brief Generation
Weekly briefs generated with executive snapshots, event timelines, narrative summaries, and investor risk overlays.
- 6Cache & Serve
Results cached to database. CDN-friendly cache headers enable fast reads with stale-while-revalidate for freshness.
Target Countries (v0.1)
24 countries selected for initial coverage based on geopolitical significance and data availability.
Update Schedule
- Pipeline Execution
- Daily at 6:00 AM IST (00:30 UTC)
- Cache TTL
- 1 hour (s-maxage=3600)
- Stale-While-Revalidate
- 24 hours
- Brief Cadence
- Weekly (ISO week format: YYYY-WNN)
Limitations & Caveats
- • v0.1 uses simulated indicator values. Production deployment requires real OSINT source integration.
- • Narrative Interference Index accuracy depends on platform API access, which may be restricted.
- • Consistency Index requires historical baseline data that may not exist for all countries.
- • Weekly resolution may miss intra-week volatility spikes.
- • Model is not designed for prediction; it measures current-state risk, not future events.
GeoPol OSINT Platform v0.1
Research / MVP-ready