Methodology
How Risk Scores Are Calculated
The Nyman Crisis Index applies machine learning techniques developed for crisis prediction and risk analysis to real-time global monitoring. Each region receives a risk score from 0-100 based on four components:
1. Event Volume (30 points)
Measures the quantity of news coverage using a logarithmic scale. More articles = higher crisis salience.
2. Source Quality (25 points)
Weights articles by source credibility. Reuters, BBC, FT = 1.0; regional specialists = 0.7; state media = 0.4; unknown sources = 0.1.
3. Conflict Intensity (25 points)
Percentage of English-language article titles containing conflict keywords: attack, strike, killed, missile, invasion, casualties, etc.
4. Sentiment Momentum (20 points)
Density of negative sentiment words (crisis, disaster, threat, escalation) in article titles, normalized per 100 words.
Data Source
All article data comes from the GDELT Project, which monitors news media in over 100 languages from 100,000+ sources worldwide. The index queries GDELT every 6 hours (00:00, 06:00, 12:00, 18:00 UTC) to capture articles within hours of publication.
Update Frequency
- 4 times daily — Overnight (00:00), Morning (06:00), Afternoon (12:00), Evening (18:00) UTC
- Real-time lag — Articles appear in the index within 6 hours of publication
- Historical data — 30-day rolling window for trend analysis
Research Foundation
This methodology builds on techniques developed through crisis monitoring and text analysis research, including work detecting conflict events and analyzing geopolitical developments in real-time data streams.
Limitations
- Reflects media coverage, not direct field intelligence
- English-language bias in sentiment/conflict scoring
- 6-hour lag from event to index update
- Does not predict future events — monitors current state only