Only information available then
Every replay is restricted to information available within its declared historical window.
Validation
How did Voluspa handle
past disruptions?
Revisit real disruptions and see what the system found using information from that time. These tests do not establish future forecasting accuracy.
Validated replay
Suez Canal · 2021
Shipping analysis
Ever Given disruption
In this historical test, Meridian identified the drop in Suez shipping traffic within a day and followed the recovery using only data available at the time.
1 day
Time to detect
71%
Average traffic drop
5 / 5
Acceptance checks
0
Early warnings
23 March–31 May 2021
View evidence →Exploratory research case
Taboneo–Banjarmasin · 2026
Meridian · observed voyage behaviour
Directional transit anomaly
Vessels departing Taboneo for inland Banjarmasin on 15–18 August showed substantially longer recorded voyage intervals than in their own earlier journeys. The cause remains unverified; this is a retrospective research case, not a validated historical replay.
25
Vessels with pre-event comparison
22 / 25
Vessels with longer intervals
+46.11 h
Median within-vessel increase
15–18 August 2026 · South Kalimantan, Indonesia
View evidence →Evaluation method
Evidence before interpretation.
Historical replays are designed to separate genuine model behaviour from retrospective storytelling.
Clear test criteria
Detection and recovery requirements are recorded before model output is evaluated.
Original results preserved
Initial runs are preserved before tuning, reinterpretation or model adjustment.
Engines tested separately
Individual modules are tested separately before integration through Yggdrasil.