Research On The Turning Point Of Stock Market Based On Behavioral Propensity Calculation

In recent decades,with the rapid development of economy,investment and financing has become an important part of family life financing.As an important financial platform of financial investment theory,the stock market attracts more and more investors and research experts' attention,and more scientific and technological methods are applied in the field of stock forecasting.The traditional stock forecasting models are basically based on the stock technical indicators,but the local characteristics of the stock technical indicators can not deeply reflect the behavior tendency of the main funds,which makes it difficult to effectively predict the stock market.Behavioral tendency refers to the idea that the subject wants to achieve a specific goal and direction,which is the root cause of decision-making.The fluctuation of stock price in stock market is the result of the operation of the main capital,and any operation of the main capital has certain behavioral intention.From the perspective of the behavioral inclination of the main capital,this paper puts forward the prediction method of the turning point of stock market based on the calculation of behavioral inclination.The specific research contents are as follows:Because the traditional piecewise linear representation method does not consider the change of data distribution in the stock market,which leads to the unreasonable segmentation,and the localization of technical indicators related to the inflexion of the stock market makes it difficult to predict the inflexion effectively,a BPC-WSVM algorithm based on behavioral propensity calculation is proposed based on the piecewise linear representation method.First of all,a piecewise linear representation(V-PLR)method of volatility distribution change is proposed to optimize the threshold value of PLR segment adaptively through volatility distribution change;then,the stock market characteristics related to the main force's behavior tendency are extracted and quantified,and the main force's behavior tendency is calculated by the fusion of the feature variables extracted by using the logic regression(LR);finally,the behavior inclination is calculated The directionality calculation results and the input characteristics of PLR-WSVM algorithm are brought into WSVM to predict the inflection point.The experimental results on real data show that the algorithm is more adaptive and the prediction accuracy is effectively improved.In view of the differences in the importance of the main behavioral tendencies reflected by different indicators,as well as the problems that high-dimensional data may contain a lot of redundant information and noise,which will bring difficulties to the extraction of effective information,etc.,based on the BPC-WSVM algorithm,a turning point prediction(PNT-WSVM)for the calculation of the importance of positive and negative tendentious features is proposed.Firstly,according to different types of inflexion(upper inflexion and lower inflexion),the proposed feature variables are classified into positive and negative categories;secondly,PCA principal component analysis is used to calculate the importance of the positive and negative index sets,extract the positive and negative principal component data sets,and then apply the two principal component data sets and the inflexion real marker to construct the data sets,which are brought into the multiclassification support vector machine.Finally,according to the inflexion type of simulation investment,experiments show that PNT-WSVM algorithm has better profitability and more stable performance.