Mean prediction model of stock return based on TSO-LSTM neural network and its application in intelligent investment
According to the historical data of stock returns,an intelligent asset allocation system driven by data and model is established to guide shareholders to maximize their returns.The LSTM neural network u-sing the TSO algorithm provides the mean and covariance matrix of the yield for the distributed Robust op-timization portfolio model,and solves the distributed Robust model more in line with the actual situation.The return of the model is significantly higher in the first 10 days than the distributed Robust model more in line with the actual situation.The return of the model is significantly higher in the first 10 days than the distributed Robust model that directly uses the historical mean,and the number of losses is less than the distributed Robust model directly using the historical mean and the average distribution of funds.Mean-while,the decision system proposed in this paper can retrain the LSTM network by updating the historical data over time,making the model maintain good results.TSO-LSTM neural network can effectively grasp the historical data characteristics of stock yield and provide investors with good investment decisions in real time.
LSTM neural networkdistributed and Robust investment portfolio optimizationtuna swarm optimization algorithmCVaR model restrain
万方数据
维普
