Contagion risk has garnered widespread attention with the major occurrence of financial crises. However, existing literature has mainly focused on the characteristics analysis of risk contagion, while ignoring the prediction of contagion risk. Therefore, considering the importance of time series charts, a framework for contagion risk prediction is proposed by comprehensively mining the features of technical charts and technical indicators. In detail, correlation and centrality of the financial network are extracted to specify the contagion risk. Then, time series charts are converted to graphs by the chart similarity, and each node in the graph is assigned with technical indicators. Finally, a Chart Graph Convolutional Network (Chart GCN) is used to extract the features in the graph and predict the contagion risk. Based on the 28 sectors of the Chinese stock market, we specify the contagion risk and verify its consistency with the occurrence of financial crisis. In addition, our model proves the effectiveness of mining time series chart features in stock market contagion risk prediction. These results are essential for risk management in the financial market.
扫码关注我们
求助内容:
应助结果提醒方式:
