Ulrich Aiounou

Aix-Marseille Université · Economics · 2026–27 job market

Fields

Econometrics, Finance

Job market paper

When Is News Really News? Evidence from Cryptocurrency Markets

This paper examines whether news sentiment provides additional predictive information for cryptocurrency returns and volatility beyond that contained in market variables. I use asset-specific news headlines collected from GDELT to construct two daily sentiment measures based on a financial lexicon and FinBERT. The analysis covers Bitcoin, Ether, Solana, and XRP, with multiple forecast horizons. Seven forecasting models are considered, including linear regressions, tree-based methods, and a neural network. Forecasting performance is evaluated using an expanding-window approach, with RMSE comparisons and the Clark-West test. I consider both a common sample and asset-native samples to examine the sensitivity of the results to the sample design. The analysis also examines whether the predictive contribution of sentiment differs between high- and low-volatility market conditions. The results indicate that news sentiment provides more favorable improvements for volatility forecasting than for returns. Statistically supported return gains are relatively limited and rarely persist across the two sample designs. In contrast, several volatility forecasts benefit from sentiment information, with the largest improvements generally obtained using Multilayer Perceptron (MLP), particularly for XRP. However, some of these gains weaken or reverse when different sample periods are considered. The results also show that improvements occur under different market volatility conditions, without a consistent advantage for either regime.