Study On The Earnings Characteristics And Forecasting Of A-share Market

In recent years,with the rapid growth and sustainable development of China's economy,the financial market has also made rapid progress and improvement.The financial industry plays an increasingly important role in our national economy,which has a great impact on economic and social stability.Therefore,it is of great significance to understand the characteristics of the stock market in a scientific and reasonable way for dispersing risks,discovering risks in time and avoiding risks as much as possible.Traditional econometric methods in China have made great achievements in income forecasting,but for the non-linear system of financial system,new methods can be studied from some new perspectives to make up for the shortcomings of traditional methods.Based on this consideration,this thesis introduces the symbolic time series analysis method which does not need to construct models and any assumptions into the financial market,and tries to analyze and study the stock returns and network structure.Firstly,this thesis introduces the research background and significance of the article,and describes the research methods,theoretical basis and structural framework applied in this thesis,and puts forward the innovative points.The first chapter is mainly about the theoretical discussion,including the process of time series symbolization,the Euclidean norm used in describing the difference between sequences and so on.Then the minimum spanning tree and the hierarchical tree method are introduced.Finally,the principle of financial income prediction is discussed.Chapter 2 mainly makes an empirical analysis on the income and network structure.Firstly,the six index earnings series of Shanghai Composite Index,Shenzhen Composite Index,Shanghai Industrial Stock,Commercial Stock,Real Estate Stock and Public Utilities Stock Index are symbolized,and the main change modes of their earnings series are obtained.Then,the differences among the series are analyzed according to the relevant statistics.Then,the earnings area is analyzed based on the main change modes.Prediction and analysis were carried out.Finally,based on Euclidean distance,the network structure of the stock market is analyzed,and the clustering effect between the underlying stocks is analyzed,and the clustering information of the underlying stocks is obtained.Chapter 3 presents a model prediction method based on sequence alignment.For a given current symbol sequence,the historical fragment with the greatest similarity is found by traversing the historical data,and its next revenue data is used to predict the next revenue value of the current sequence.The prediction effect is evaluated by calculating the MAPE value,which also shows the feasibility of this method.Then the static and dynamic methods are used to symbolize the Shanghai Composite Index,and the results are compared to get the optimal method.Through the analysis of the characteristics of financial returns,the main change patterns of the earnings level of the Shanghai Composite Index and other earnings series studied in this thesis are that the four consecutive trading days are medium returns and the four consecutive trading days are medium returns,medium returns,low returns and high returns.Through the analysis of the differences among the indexes,it is concluded that the correlation between the Shanghai Industrial Stock Index and the real estate stock index is relatively small.The income difference is large,the correlation between Shanghai real estate stock index and public utility stock index is large,and the income difference is the smallest.Subdividing into individual constituent stocks,we can see that the studied constituent stocks have the characteristics of industry clustering,region clustering,cooperative distribution clustering and shareholder distribution clustering,that is,the stock price is highly correlated,which has a strong guiding significance for investors to diversify risk.In the revenue forecasting based on sequence alignment,the MAPE values are less than 0.1,which shows that the sequence alignment method has better forecasting effect.By comparing the prediction accuracy between the dynamic method and the static method,it is found that the dynamic symbolization method is superior to the static method,which is consistent with the research results of other scholars.