Forecasting realized volatility using news flow

Marcelo Fernandes

doi:10.1016/j.qref.2025.102040

Forecasting realized volatility using news flow

2025-09-04

0

PRE

AI

DOI：10.1016/j.qref.2025.102040

Original

Original request for help

Save

Abstract

En 中文

• We build firm-specific and economic-policy uncertainty indicators from a corpus of roughly 400,000 economic-related news articles. • The news-based indicators provide significant predictive gains for realized volatility, even after controlling for key empirical features. • Forecasting gains are most pronounced for highly liquid stocks and multi-day horizons. • Regularization-based model selection identifies news flow and lagged negative returns as the most informative predictors.

Forecasting realized volatility using news flow

Abstract

Journal

Organization