| The stock market plays an important role in the securities industry and the financial industry.The stock market is a “barometer” of the macro economy and the mood of investors.The economic development of any country is inseparable from the stock market.Since the establishment of the Shanghai Stock Exchange in 1990,the Chinese stock market has reached a scale that can be achieved in many countries for decades or even hundreds of years in just 29 years.Effective stock forecasting is the key to the success of the investment for enterprises and investors.The more accurate the forecast,the more sure the risk prevention will be;the economic development of the country is important for macroeconomic policy formulation.Indicators,the more accurate the forecast,the stronger the ability to regulate the market risk.Therefore,it is especially important to correctly analyze and forecast the stock price,but because the stock is subject to many complicated factors such as social,political,economic,cultural and investor psychology.The impact,coupled with the uncertainty of these influencing factors,makes it difficult to predict stock prices.As a core theory of behavioral finance,limited attention is considered.In the case of many complicated information,investors are unable to obtain and fully understand information in time due to limited time and energy.Investors are more inclined to purchase their own concerns.stock.With the advent of Internet search engines,investors use search engines to search for information they care about as a popular choice.At the same time,investors’ search for relevant information will be recorded by search engines,which directly measures investors.The attention has provided accurate data and also offers the possibility to use the investor’s attention to predict the stock market.As a set of time series,the stock price is nonlinear and dynamic,which makes scholars begin to transform from traditional time series analysis models to artificial analysis.Recurrent Neural Network(RNN)is a neural network used in time series prediction in machine learning,and Long Short-Term Memory(LSTM)as a variant of circulating neural network.It is more adept at exploring long-term dependencies between time series data and is therefore more suitable for stock price forecasting.This paper builds a keyword lexicon based on Baidu index,and uses LSTM deep neural network to model and forecast the ups and downs of Shanghai Composite Index with its search volume as the investor attention index.In the research,this paper uses LSTM deep neural network to predict the Shanghai Composite Index.The results show that the investor attention index based on Baidu index has a good predictive effect on the Shanghai Stock Index.Secondly,the accuracy of the LSTM model is further improved by the selection of different optimization methods and different feature keywords.Furthermore,through empirical analysis,it is found that the investor attention index based on the Baidu index is a synchronous indicator,and investors are concerned that the change is consistent with the change of the Shanghai Composite Index.Finally,by comparison,the use of comprehensive search volume(pc end + mobile end)to predict the rise and fall of the Shanghai Composite Index is higher than the use of PC or mobile search data alone.This paper constructs the investor attention index method and further innovates on the basis of predecessors.It constructs relevant keyword database through Baidu search recommendation engine,and uses random forest algorithm to sort the importance of related keywords,and takes relevant keywords with cumulative contribution of 80%.The search volume builds the investor attention indicator;in addition,the LSTM algorithm prediction model is further optimized to improve its accuracy.Therefore,the results of the above empirical research have certain reference significance for both the investment decision of individual investors and the formulation of the system by the market supervision department. |
PDF Full Text Request