Scientists develop a forecasting system that predicts high-risk windows and regions for solar superflares, using 50 years of X-ray data and machine learning techniques.