Stock Trend Forecasting Method Based On Intrinsic Attributes Of Stocks

With the reform and opening up,China’s economy has developed rapidly and occupies an increasingly important position in the global economy.At the same time,the stock market has become more and more important in the national economy,and how to accurately predict the stock market has become the focus of many scholars’ research in recent years,which has rich social and economic value,but also has important academic value.In recent years,with deep learning has made amazing progress in many fields,such as computer vision,natural language processing and so on.More and more scholars have introduced it into stock market forecasting,built many effective models and made good progress.These models,however,bring significant im-provements over traditional statistical models.But for now,the advantage of sub-jective trading is still great,because the experts of subjective trading have an understanding of the intrinsic properties of stocks,which provides a wealth of information for stock investment.Often experienced investors will divide stocks into different categories according to their different intrinsic attributes.Accord-ing to different attributes,they have different investment principles for different stocks.For example,for stocks with low volatility,experts tend to hold them for the long term,while for cyclical stocks,professional investors tend to buy them at the bottom of the business cycle and sell them at the top,a more frequent trading strategy.In addition.in the stock market,we can usually observe such a phenomenon.in the same trading day,often have similar properties with the same rise and fall situation.For example.one trading day,liquor plate stocks rose at the same time,another trading day.the concept of blue chip stocks have a collective drop,while the small-cap stocks have a collective rise.Investors with rich experience can take advantage of the intrinsic properties of stocks to buy a large number of stocks with liquor properties in the liquor market,and buy stocks with small market value properties in the medium and small-cap market,so as to obtain high returns.The intrinsic properties of such stocks play a very important role in stork in-vestment decisions,but it is very difficult to obtain such information,and it is also very abstract.Although we can ask experienced investors to manually label,but this method is very expensive.To this end,this paper proposes the assump-tion that stocks with similar intrinsic attributes have similar market performance,directly from the stock market performance,the construction of stock rise and fall co-occurrence matrix,and the derivation and establishment of a mathemat-ical model,the extraction of the potential representation vector of the stock,as the intrinsic attributes of the characteristic representation.At the same time,in order to strengthen the relationship between the intrinsic attributes of stocks and market sentiment,the establishment method of market sentiment preference and market sentiment aversion is constructed to enhance the dynamic correlation between individual stocks and the market.At the same time,based on the traditional deep learning stock market pre-diction model,a new deep learning framework is proposed.which can effectively use the concept of intrinsic attributes of stocks and market sentiment preference proposed in this paper,so as to achieve more accurate stock prediction.In this paper,the historical data of stock market of CSI 800 index are used to conduct experiments on stock price regression,stock rise and fall classification,and stock return ranking respectively.The results are all better than the benchmark model,which verifies the validity of the model proposed in this paper in the task of stock prediction.