Scholarlink

With huge risks in financial investment, how to efficiently transform from subjective investment to scientific quantitative investment is particularly crucial. For this reason, this paper proposes a mean-CVaR portfolio model based on graph neural network (GNN), aiming at improving the scientificity and efficiency of investment decisions. The variance of future stock returns is predicted by graph convolution network and compared with XGBoost and LSTM models. Then, this research further uses CVaR to quantify investment risk and builds a financial portfolio model based on mean-CVaR. The stock data in the field of new energy vehicles are selected for prediction and analysis to solve the optimal investment portfolio. The research results confirm the feasibility and superiority of the model in portfolio management.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mean-CVaR Portfolio Model Based on Graph Neural Network

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Guogang Zheng

摘要

With huge risks in financial investment, how to efficiently transform from subjective investment to scientific quantitative investment is particularly crucial. For this reason, this paper proposes a mean-CVaR portfolio model based on graph neural network (GNN), aiming at improving the scientificity and efficiency of investment decisions. The variance of future stock returns is predicted by graph convolution network and compared with XGBoost and LSTM models. Then, this research further uses CVaR to quantify investment risk and builds a financial portfolio model based on mean-CVaR. The stock data in the field of new energy vehicles are selected for prediction and analysis to solve the optimal investment portfolio. The research results confirm the feasibility and superiority of the model in portfolio management.