Renewable energy stocks forecast using Twitter investor sentiment and deep learning

Abstract

This paper examines the impact of investor sentiment on forecasting returns and volatility for renewable energy stocks. We apply a natural language processing technique to extract investor sentiment from Twitter during both trading and non-trading hours. Forecasting analyses are conducted using a state-of-the-art hybrid deep learning technique and benchmark models. Results show that the sentiment variables hold significant add-on information not captured by standard financial market variables. Twitter investor sentiment considerably improves return and volatility forecasts of renewable energy stocks, especially when the deep learning method is employed. Our results are statistically significant and robust under different settings.