Research And Implementation Of Portfolio Analysis System Based On Deep Reinforcement Learning

Intelligent and efficient portfolio analysis is of great significance for individual investors to judge the future trends of the financial market,adjust portfolio in a timely manner,avoid investment risks,and increase investment returns.Due to the complex,dynamic and unstructured characteristics of financial data,it is difficult to discover the laws of financial market operation using traditional data analysis methods.Therefore,it is of great research value to provide effective investment advice to individual investors with the help of the deep reinforcement learning algorithm's powerful data processing and analysis capabilities.Firstly,this paper consults a large number of domestic and foreign research documents and finds that the use of factor index data in the stock selection process needs to be improved,and the application of deep reinforcement learning algorithms in the field of portfolio analysis is relatively small.For the above situation,this paper proposes a multi-factor stock selection model based on improved LSTM deep neural network for stock selection in the portfolio.For the input of factor index data,this paper improves the LSTM deep neural network to process factor data and effectively improves the utilization rate of multiple factor data.Through the analysis of multiple experiments,the Adam model optimizer was selected to optimize the prediction accuracy of the multi-factor stock selection model.Secondly,this paper attempts to apply deep reinforcement learning algorithms to portfolio analysis,constructs a deep reinforcement learning model to simulate market transactions,and realizes the trading signal prediction of each asset in the portfolio for a period of time in the future.It is convenient for individual investors to adjust the weight of corresponding assets in a timely manner,reduce the risk of investment portfolios and increase returns.In order to more realistically simulate the market environment,this paper adds the news data of the corresponding assets to the deep reinforcement learning model.Through experiments,it is found that it has a good effect on balancing the risk and return of the portfolio.Finally,in order to improve the user experience of individual investors,this paper designs and develops a portfolio analysis system,applying the deep reinforcement learning algorithm in the experiment process to the system.The system can provide functions such as multi-factor stock selection,portfolio asset selection and asset allocation,portfolio return and risk analysis.This is of great significance to the application and practice of deep reinforcement learning algorithms in the field of portfolio.