| The problem of stock price prediction is a very important research topic in financial market research.Predicting the future market situation of the stock market can guide investment behavior and help investors to allocate funds reasonably.In the stock price prediction problem,the traditional research methods are mainly based on various time series analysis models in econometrics,and then the traditional machine learning models represented by SVM and random forest have been applied to the stock price prediction problem.Traditional research methods mainly focus on the impact of individual stock historical information on stock prices,without considering the linkage effect between stocks.In recent years,with the development of artificial intelligence technology,the intelligent model represented by deep learning has improved the traditional method and showed better performance in the stock market.At the same time,the linkage effect between stocks has gradually been studied by more researchers focus on.Existing research shows that there is a linkage effect in my country’s stock market,in which industry correlation is the fundamental reason for the long-term linkage of stock prices,and market behavior(such as supply and demand,herd effect,information spillover,etc.)is an important reason for the short-term linkage effect of stock prices.At present,the use of deep learning methods such as graph neural networks to study and model these two linkage effects is relatively rare.At the same time,behavioral finance believes that investor sentiment has a significant impact on the stock price linkage effect.As professional investors with good investment and research capabilities,fund managers will make judgments on future industry trends based on their investment experience combined with the current market environment.These views and emotions are often more forward-looking and accurate.However,there are basically no researchers in the existing literature that incorporate the industry sentiment information of fund managers into the stock price prediction model,and the research on the impact mechanism of the industry sentiment of fund managers on the stock linkage effect is also relatively small.Based on the industry linkage effect of the stock market and the industry sentiment of public fund managers,this paper proposes a new stock price prediction framework that integrates multi-source information.The main tasks are as follows:(1)This paper considers the growing influence of public fund managers in the stock market,and extracts the industry position information of fund managers by performing multiple regression on the fund’s net value return sequence and the industry index sequence.Combining the industry sentiment data of fund managers with the stock relationship graph data under the industry association,the long-term linkage effect of stocks is mined.(2)This paper considers the short-term linkage effect of the industry,and proposes an attention-based inter-industry attention layer to learn the dynamic embedded representation of the industry to capture the short-term linkage relationship between industries.(3)This paper proposes a new stock price prediction model structure,which integrates multi-source data such as traditional volume and price data,fund manager industry sentiment,and stock relationship graphs to capture the long-term and shortterm linkage effects between stocks,thereby improving stock price prediction.performance of the model.In order to verify the effectiveness of the method proposed in this paper,this paper designs several models and conducts comparative experiments on the data set.The experimental results show that the method proposed in this paper has a certain improvement compared with the classic stock price prediction model in terms of IC value,Rank IC,Precision and other evaluation indicators.In order to verify the effectiveness of each newly added module,we designed a comparative experiment to verify the effectiveness of each module.We found that when the industry sentiment information of fund managers or the long-term and short-term linkage effect modules of stocks were removed,the predictive performance of the model decreased significantly.The experimental results verify the effectiveness of the stock price prediction model based on the long-term and short-term linkage effect fusion of the industry proposed in this paper. |
