Methodology
How ATLAS builds a daily country score.
ATLAS explains exposure with a two-layer model: a structural baseline for every country and a capped live layer for recent movement.
The score is an exposure index, not a strike prediction, and quiet news windows keep the dynamic layer low rather than removing the country from the model.
Score formula
Baseline Score
0-70
Always-on structural exposure for every country.
Dynamic Score
0-30
Recent movement from news and verified pressure.
ATLAS Score
0-100
The daily total exposure score shown across the product.
Two-layer model
Baseline 0-70 + Dynamic 0-30
Coverage rule
Every listed country keeps a valid score
Interpretation
Score and confidence answer different questions
Latest publication: Apr 2, 2026, 14:00 UTC
Model version 2026.04.03Baseline vs dynamic
The 70 / 30 structure at a glance
Baseline explains persistent exposure. Dynamic explains current movement. The component bars below show how each layer is weighted.
Baseline Score
0-70
70 points
A structural layer that exists for every country, including low-news states and microstates.
This is where persistent exposure lives: geography, alliances, flashpoints, and state presence.
Nuclear actor adjacency / exposure
Persistent exposure to nuclear-armed states and nearby strategic rivalry.
Alliance entanglement
Treaty posture, alliance pull, and indirect crisis linkage.
Regional flashpoint proximity
Distance to active or persistent escalation theaters.
Strategic relevance
Strategic position, chokepoints, and broader geopolitical centrality.
Civilian / public exposure context
Civil infrastructure and public exposure as supporting context.
Baseline floor / state presence
A minimum structural floor so quiet countries still remain in the index.
Dynamic Score
0-30
30 points
A recent-movement layer that rises when current reporting or verified pressure materially changes.
Reuters-linked coverage informs the rhetoric slice today, while verified event inputs can expand through the dedicated event layer.
Verified conflict / event pressure
The direct event layer reserved for verified conflict pressure.
News / rhetoric escalation
Recent Reuters-linked coverage and public escalation language.
Sudden movement / cluster severity
Sharp movement clusters that accelerate a quiet baseline.
Example score card
One country, broken into the same parts users see in ATLAS
This is an illustrative breakdown showing how a country can carry a meaningful baseline and a smaller recent-movement lift at the same time.
Illustrative country score
Poland
Baseline
43.0
of 70
Dynamic
9.0
of 30
Total
52.0
of 100
Structural exposure sets most of the score, while current reporting adds movement on top rather than defining the country from scratch.
Top drivers
The example score is explained in plain language, with the layer called out directly.
Alliance entanglement
BaselineTreaty posture and forward alignment keep the structural floor elevated.
Regional flashpoint proximity
BaselineEastern European proximity keeps persistent exposure in the model.
Reuters-linked escalation coverage
DynamicRecent rhetoric and movement lift the score without replacing the baseline.
Low-news countries
Quiet news no longer means score absence
The old logic depended too heavily on publication volume. The new logic keeps the country in the index and lets the dynamic layer fall to zero when the window is quiet.
Old logic
No recent news
->
No score
Publication volume could determine whether a country showed up at all.
New ATLAS logic
Fiji example
Baseline
11.0
of 70
Dynamic
0.0
of 30
Total
11.0
valid low score
The baseline keeps a quiet country in the index, and the dynamic layer can stay at zero without collapsing the score.
Score vs confidence
Exposure level and interpretive strength are separate
Score answers how exposed the country looks. Confidence answers how strongly the current evidence supports that reading.
Score = exposure level
52.0
out of 100
Score measures how much exposure ATLAS sees after combining structure and current movement.
Confidence = interpretive strength
Medium
Confidence measures how firmly the model can interpret that score from the available evidence.
A country can have a high score with only medium confidence if the structure is clear but the latest reporting is thin or uneven.
Important notes
Read the score as structured context, not prediction
Not a probability
A score of 80 does not mean an 80 percent chance of nuclear use.
Public-source model
ATLAS does not include classified intelligence or closed military planning.
Mixed update speeds
Baseline inputs move slowly, while rhetoric and event layers can move daily.
Directional reading
Use score, movement, and confidence together rather than any one field alone.