Research preview

The sudden deployment of Eritrean forces in Ethiopia’s northern Tigray region has revived concerns that a localized civil war could spread across the Horn of Africa. For quantitative traders, such geopolitical shocks create abrupt movements in commodity prices, sovereign spreads, and risk‑off assets. This article shows how high‑dimensional machine‑learning tools can turn sparse news feeds into actionable signals while guarding against over‑fitting. From News to Numeric Features The first step is to translate unstructured headlines into a structured predictor matrix. A simple pipeline extracts the date, location, actor (e.g., “Eritrean troops”), action (e.g., “deployed”), and conflict intensity tags (low, medium, high). Each element becomes a binary or categorical variable....