The Return Direction Predict Model Based On Time-varying Probabality Density Function

In modern times under the impetus of the development of the entire financial industry,the stock market and the securities industry in our country,more and more get the attention of many investors,more and more people are willing to join the stock market and financial market,the investment behavior of investors and their expected earnings,well this in great extent promoted the prosperity of the Chinese stock market.In their investment behavior,many investors are starting to realize the importance of the stock market direction of directional prediction.Therefore,the return on the stock market direction change the depth of the analysis and prediction has great economic significance and value of practical application.Scholars at home and abroad for the development of the stock market,also reached a certain level.With the continuous development of information technology,the new technical analysis and theoretical knowledge are constantly being injected into the mathematical model.Because our country financial industry gradually standardized,as well as the domestic and foreign scholars and investors for a detailed analysis and the urgent need of the direction of the stock market are the driving force to the development of the paper.The main purpose of this paper is to predict the rate of return on the stock market,that the direction of the stock return rate to a certain extent,is predictable.The direction of stock return forecast is:based on the historical data of the stock market and its historical trend,in the next period of time to the stock market to predict the direction of the rate of return.Based on Harvey and Oryshchenko(2012)proposed time-varying probability density function theory of the application and expansion of the use of non-parametric model of the stock market rate of return for the direction of prediction.The advantage of nonparametric models is that their data do not need to satisfy certain distributions.Because the sample data in this paper is the stock return rate data,it belongs to the time series data,and the time series data will be affected by many accidental factors,thus showing the randomness.In this paper,we use the nonparametric model based on the time-varying probability density function to predict the probability of the direction of the next stock return in the historical data information set of the stock market.By using the relationship between the direction prediction and the second order moment,Direction forecasting probability adjustment mechanism,which is the innovation of this paper.Finally,a binary selection model is introduced as a contrastive model.Empirical research on Chinese stock market data shows that our benchmarking model and adjustment mechanism based on the time-varying probability density function have a significant out-of-sample forecast for the direction of stock market returns both in the statistical sense and in the economic sense ability.Moreover,the adjustment mechanism also shows a better predictive power than the yield-based forecasting baseline model.Finally,compared with the binary selection model,both the prediction model of the yield direction and its adjustment mechanism are better than the binary selection model.