Research

Identifying recessions using machine learning

Knowledge of whether a recession is imminent provides valuable information for investors to position their portfolios.  We use Support Vector Machines (SVM), a machine learning algorithm, to identify recessions in real time.  The SVM model shows excellent performance for this task, with similar or more accurate nowcasts compared to existing models, while providing more timely estimates.  The important advantage of timely estimates makes SVM more useful for tactical asset allocation.