Country Instability Index

Real-time stability scoring for 24 strategically significant countries, combining unrest, security, and information signals into a dynamic instability metric.

The Country Instability Index maintains a real-time instability score for strategically significant countries. Rather than relying on static risk ratings, the CII dynamically reflects current conditions based on multiple input streams.

Monitored Countries (Tier 1)

Region	Countries
Americas	United States, Venezuela, Brazil, Mexico, Cuba
Europe	Germany, France, United Kingdom, Poland
Eastern Europe	Russia, Ukraine
Middle East	Iran, Israel, Saudi Arabia, United Arab Emirates, Turkey, Syria, Yemen
Asia-Pacific	China, Taiwan, North Korea, India, Pakistan, Myanmar

Three Component Scores

Each country’s CII is computed from three weighted components:

Component	Weight	Data Sources	What It Measures
Unrest	40%	ACLED protests, GDELT events	Civil unrest intensity, fatalities, event severity
Security	30%	Military flights, naval vessels	Unusual military activity patterns
Information	30%	News velocity, alert clusters	Media attention intensity and acceleration

Scoring Algorithm

Unrest Score:
  base = min(50, protest_count × 8)
  fatality_boost = min(30, total_fatalities × 5)
  severity_boost = min(20, high_severity_count × 10)
  unrest = min(100, base + fatality_boost + severity_boost)

Security Score:
  flight_score = min(50, military_flights × 3)
  vessel_score = min(30, naval_vessels × 5)
  security = min(100, flight_score + vessel_score)

Information Score:
  base = min(40, news_count × 5)
  velocity_boost = min(40, avg_velocity × 10)
  alert_boost = 20 if any_alert else 0
  information = min(100, base + velocity_boost + alert_boost)

Final CII = round(unrest × 0.4 + security × 0.3 + information × 0.3)

Scoring Bias Prevention

Raw news volume creates a natural bias: English-language media generates far more coverage of the US, UK, and Western Europe than conflict zones. Without correction, stable democracies would consistently score higher than actual crisis regions.

Log Scaling for High-Volume Countries

Countries with high media coverage receive logarithmic dampening on their unrest and information scores:

if (newsVolume > threshold):
  dampingFactor = 1 / (1 + log10(newsVolume / threshold))
  score = rawScore × dampingFactor

This ensures the US receiving 50 news mentions about routine political activity doesn’t outscore Ukraine with 10 mentions about active combat.

Conflict Zone Floor Scores

Active conflict zones have minimum score floors that prevent them from appearing stable during data gaps or low-coverage periods:

Country	Floor	Rationale
Ukraine	55	Active war with Russia
Syria	50	Ongoing civil war
Yemen	50	Ongoing civil war
Myanmar	45	Military coup, civil conflict
Israel	45	Active Gaza conflict

The floor applies after the standard calculation. If the computed score exceeds the floor, the computed score is used. This prevents false “all clear” signals while preserving sensitivity to actual escalations.

Instability Levels

Level	Score Range	Visual	Meaning
Critical	81-100	Red	Active crisis or major escalation
High	66-80	Orange	Significant instability requiring close monitoring
Elevated	51-65	Yellow	Above-normal activity patterns
Normal	31-50	Gray	Baseline geopolitical activity
Low	0-30	Green	Unusually quiet period

Trend Detection

The CII tracks 24-hour changes to identify trajectory:

• Rising: Score increased by ≥5 points (escalating situation)
• Stable: Change within ±5 points (steady state)
• Falling: Score decreased by ≥5 points (de-escalation)

Contextual Score Boosts

Beyond the base component scores, several contextual factors can boost a country’s CII score (up to a combined maximum of 23 additional points):

Boost Type	Max Points	Condition	Purpose
Hotspot Activity	10	Events near defined hotspots	Captures localized escalation
News Urgency	5	Information component ≥50	High media attention indicator
Focal Point	8	AI focal point detection on country	Multi-source convergence indicator

Hotspot Boost Calculation:

• Hotspot activity (0-100) scaled by 1.5x then capped at 10
• Zero boost for countries with no associated hotspot activity

News Urgency Boost Tiers:

• Information ≥70: +5 points
• Information ≥50: +3 points
• Information <50: +0 points

Focal Point Boost Tiers:

• Critical urgency: +8 points
• Elevated urgency: +4 points
• Normal urgency: +0 points

These boosts are designed to elevate scores only when corroborating evidence exists. A country must have both high base scores AND contextual signals to reach extreme levels.

See Geographic Convergence for the full algorithm.

Server-Side Pre-Computation

To eliminate the “cold start” problem where new users would see blank data during the Learning Mode warmup, CII scores are pre-computed server-side via the /api/risk-scores endpoint. See the Server-Side Risk Score API section for details.

Learning Mode (15-Minute Warmup)

On dashboard startup, the CII system enters Learning Mode, a 15-minute calibration period where scores are calculated but alerts are suppressed. This prevents the flood of false-positive alerts that would otherwise occur as the system establishes baseline values.

Note: Server-side pre-computation now provides immediate scores to new users. Learning Mode primarily affects client-side dynamic adjustments and alert generation rather than initial score display.

Why 15 minutes? Real-world testing showed that CII scores stabilize after approximately 10-20 minutes of data collection. The 15-minute window provides sufficient time for:

• Multiple refresh cycles across all data sources
• Trend detection to establish direction (rising/stable/falling)
• Cross-source correlation to normalize bias

Visual Indicators

During Learning Mode, the dashboard provides clear visual feedback:

Location	Indicator
CII Panel	Yellow banner with progress bar and countdown timer
Strategic Risk Overview	”Learning Mode - Xm until reliable” status
Score Display	Scores shown at 60% opacity (dimmed)

Behavior

Minutes 0-15: Learning Mode Active
  - CII scores calculated and displayed (dimmed)
  - Trend detection active (stores baseline)
  - All CII-related alerts suppressed
  - Progress bar fills as time elapses

After 15 minutes: Learning Complete
  - Full opacity scores
  - Alert generation enabled (threshold ≥10 point change)
  - "All data sources active" status shown

This ensures users understand that early scores are provisional while preventing alert fatigue during the calibration period.

Keyword Attribution

Countries are matched to data via keyword lists:

• Russia: russia, moscow, kremlin, putin
• China: china, beijing, xi jinping, prc
• Taiwan: taiwan, taipei

This enables attribution of news and events to specific countries even when formal country codes aren’t present in the source data.