| With the continuous development of the Internet,stock volatility is increasingly affected by Internet information.Investors will use the Internet to read relevant stock news,understand the development trend of related industries,and even exchange information through the network.Therefore,traditional stock market volatility prediction methods that rely on stock quantitative trading are difficult to make full use of effective market information,and considering the influence of Internet information can help improve stock market volatility prediction performance.This paper focuses on multiple sources in the field of stock prediction.The data fusion problem is studied,and a stock volatility prediction model based on broad learning is proposed to improve stock prediction accuracy.The stock prediction framework proposed in this paper is based on the coupled hidden Markov model.On the one hand,the model combines stock quantitative information and stock news event information.On the one hand,it considers the relationship between stocks,which can effectively suppress the problem of data sparseness.In addition,in order to further utilize the relationship between stocks and the relationship of stock price changes in time series,this paper also proposes a Two-dimensional algorithm based on time and space,and further corrects the results of coupled hidden Markov models.Then,in order to solve the problem that the calculation of the probability graph model is slow and the calculation speed is slow,a local approximation method based on LSTM is proposed,which can effectively improve the prediction speed.At the same time,this paper proposes a maximum consensus prediction method based on bipartite graphs,which can improve the accuracy of the prediction method based on likelihood values.Finally,this paper uses the 2015 data of CS100 to verify the effect on the framework proposed in this paper.The results show that our method can effectively improve the accuracy of model prediction. |
PDF Full Text Request