Global | Geopolitics, geoeconomics and risk: a machine learning approach

Summary

We introduce a novel high-frequency daily panel dataset of both markets and news-based indicators for 42 countries across both emerging and developed markets.

Key points

• Key points:
• We study how sentiment affects sovereign risk (CDS spreads) and its forecasting power versus traditional drivers like monetary policy and volatility.
• Our analysis shows that adding news-based indicators improves forecasting accuracy, with nonlinear methods like Random Forests giving the biggest gains.

Geographies

Topics

Tags