Font Size: a A A

Research On Share Pledge Risk Prediction Based On Artificial Intelligence

Posted on:2023-07-18Degree:MasterType:Thesis
Country:ChinaCandidate:X Y CuiFull Text:PDF
GTID:2569307061955329Subject:Finance
Abstract/Summary:PDF Full Text Request
With the continuous development of share pledge,many problems and potential risks are gradually exposed.Especially since 2018,the risk of share pledge has been overstocked with the substantial price fluctuation of the stock market,the increasing market value of pledged shares of major shareholders and the trend of tight supervision.These potential risks will have a serious impact on the stability of the financial market.It is of great significance for us to supervise and resolve the share pledge risk.In recent years,machine learning has developed rapidly.It has been used in various fields widely,including the financial field,and has achieved many ideal results.Model is built on the basis of LSTM to predict risk of share pledge.In terms of input features,many factors are selected,including enterprise finance and share pledge.Risk indicator is established with the volatility rate of market price,pledge rate and closing line and is used as a label.FCN algorithm is added in LSTM to build a LSTM-FCN hybrid model.It can extract local features and temporal sequential characteristics at the same time.Accuracy,F1 score,recall,precision and hamming distance are selected as evaluation indexes because the model belongs to multiclass classification.Parameters can realize optimization according to the evaluation results.Results are compared with other models.The prediction performance of LSTM-FCN is ideal and better than other comparative models.Among all characteristics,those related to share pledge and enterprise financial characteristics of solvency,profitability and growth are more important than others.Machine learning is applied to the field of share pledge and helps to establish a risk prediction model,which has achieved ideal results.The research results can be used as a reference for the risk supervision of share pledge.Firstly,technology and advanced algorithms such as machine learning can be used to establish and improve the risk monitoring and early warning system,so as to improve the infrastructure of risk supervision.Secondly,the supervision of pledged shares and pledge proportion should be strengthened so as to enlarge the scope of supervision.Thirdly,the supervision and regulation of share pledge also need to be enhanced by improving institutional mechanisms,including differentiated supervision of risk level,dynamic risk monitoring and information disclosure.
Keywords/Search Tags:Share pledge, Machine learning, Long Short Term Memory
PDF Full Text Request
Related items