Prediction of systemic risk of China's financial industry based on phase space reconstruction and machine learning

Forecasting systemic financial risks is important to prevent and resolve systemic financial risks effectively. In this paper, we give a method to measure and predict the systemic financial risk of China’s financial industry based on stock trading data of 36 China’s A-share listed financial institutions. We establish a complex financial network and obtain three clusters of the system by spectral clustering. A weighted average value is defined to measure the systemic risk of each cluster. Based on the interrelationship of different clusters, phase space reconstruction and delay parameterization method (RDPM), radial basis function neural network (RBF) are used together to forecast the systemic financial risk. The relative accuracy of prediction is more than 95%, which is higher than that of some common machine learning methods.