Methodology v1.0 · bumped whenever the weights, baseline, signals or bands below change.
The index turns a daily, structured reading of open-source reporting into one number from 0 to 100, measuring tension relative to recent weeks. It is not a probability of conflict and not a forecast.
Each day an automated, AI-assisted analysis reviews open-source reporting and scores four categories from 0–10 against a fixed, published rubric. The scores, weights and formula below are exactly what produces the published number — nothing is adjusted by hand afterwards.
| Category | Weight |
|---|---|
| Military posture | 30% |
| Diplomatic / political | 25% |
| Economic / informational | 15% |
| Rule of law & institutions | 30% |
1. Baseline. For each category, the average and spread of its scores over the previous 30 days (at least 7 needed).
2. Deviation. How far today's score is from that baseline, in standard deviations (σ), combined across categories using the same weights.
3. Risk index = Level + 8.0 × deviation (deviation capped between -2.0σ and +3.0σ), limited to 0–100.
A category "spikes" at ≥1.5σ above its own baseline. 2+ spiking within 3 days is a "converging" signal; 3+ is "compound" — several kinds of evidence moving together is a stronger warning than any one category jumping alone.
| Band | Range |
|---|---|
| Calm | 0–24 |
| Elevated | 25–49 |
| Tense | 50–74 |
| Crisis | 75–100 |
This page is a short summary for search engines and quick reference. The full interactive methodology page has a worked example from today's reading, the reasoning behind each weight, the indicator matrix, and the changelog of past revisions. See also the FAQ and the data's CC BY 4.0 licence.