Findings from a Prediction Challenge to Forecast Conflict with Uncertainty

• In Event: Political Forecasting Methods: Challenges and Potential Solutions

Abstract

Governmental and non-governmental organizations have increasingly relied on early-warning systems of conflict to support their decision making. Predictions of war intensity as probability distributions prove closer to what policy makers need than point estimates, as they encompass useful representations of both the most likely outcome, and the lower-probability risk that conflicts escalate catastrophically. Point-estimate predictions, by contrast, fail to represent the inherent uncertainty in the distribution of conflict fatalities. Yet, current early warning systems are preponderantly focused on providing point estimates, while efforts to forecast conflict fatalities as a probability distribution remain sparse. Building on the predecessor VIEWS competition, we organize a prediction challenge to encourage endeavours in this direction. We invite researchers across multiple disciplinary fields, from conflict studies to computer science, to forecast conflict intensity as a probability distribution over the outcome, for both a true future and a test set partition, and for two levels of analysis: country-month, and PRIO-GRID-month. This article introduces the goal and motivation behind the prediction challenge, presents a set of evaluation metrics to assess the performance of the forecasting models, describes the benchmark models which the contributions are evaluated against, and summarises the salient features of the submitted contributions. The true future of the competition will be ongoing by September – the participants will submit forecasts for July 2024-June 2025 in June 2024.

Authors

• Paola Vesco, Peace Research Institute Oslo (PRIO)

• Michael P. Colaresi, University of Pittsburgh

• Jonas Vestby, PRIO

• Håvard Hegre, Peace Research Institute Oslo