The ACLED Conflict Alert System (CAST) forecasts organized political violence up to six months into the future. Forecasts are made globally - at the national and sub-national levels - and can be disaggregated by event type (battles, explosions/remote violence, or violence against civilians). Every month, ACLED CAST produces a report highlighting key conflict developments and significant predictions for the coming months.

This report reflects the May 2024 update of global forecasts.

May Global Forecast

Total Forecast: 12,189 events

Forecasted 10 Most Violent Countries

May Early Warning Watch List

The maps show CAST’s total organized violence prediction at the ADMIN1 level for May 2024. Hover to reveal forecasted event counts.

The charts depict the proportion of each event type of the total May 2024 forecast. Hover to see event totals within each event type.

The tables show the top 5 most active ADMIN1s within each country, with values denoting total organized violence.

Russia and Ukraine

Israel, Lebanon, and Palestine

Syria

Mexico

Sudan

April Global Performace

Global Accuracy, by Event Type

Global Accuracy

CAST predicted 12,218 organized political violence events for April. There were ultimately 273 fewer events than predicted, a 2% difference. This difference was driven primarily by an increase of 349 Battles events relative to the forecast.


High Activity Countries within 5% Accuracy

Accuracy Highlights

Of the countries and territories with at least 20 violent events in April, four were within 5% of their forecast.

Explore the Accuracy Metrics tab of the CAST dashboard for performance information across the world.


Large Deviations from Forecasts

Significant Changes from Forecasts

CAST under-predicted total events in two of the five largest deviations from the April forecast (Mexico, Sudan), while the remaining three largest deviations were from over-prediction (Russia, Syria, Palestine)

Track evolving conflict dynamics in these countries and others around the world each week with ACLED’s Trendfinder.