Beer, Cars & Fundamentals: Predicting German M& A activity

This paper introduces a predictive model for German mergers and acquisitions (M& A) activity leveraging deep feedforward neural networks (DFNN) incorporating well-established traditional variables (also known as features), along with a ChatGPT-based M& A sentiment score (MASS) and unconventional predictors such as beer sales and weather data. We demonstrate that the inclusion of sentiment and non-traditional variables enhances predictive performance. Our findings provide an important empirical foundation for understanding near-term fluctuations in German M& A activity and offer a forecasting tool relevant to both practitioners and researchers.