Research And Application Of Quantitative Investment Selecting Stock Model

Quantitative investment is the use of mathematical models and computer programs to predict the trend of stock prices.Because of its own objectivity,rapidity,discipline,and other advantages are increasingly favored by investors.Moreover,in the not fully matured Chinese market,there are more market information asymmetries and market failures.At the same time,the investment philosophy of most investors is relatively backward and the level of investment is also uneven.Therefore,the study of quantitative investment models has important guiding significance for the adjustment of national macroeconomic policies and personal investment.The prediction of the stock price trend can be studied as a classification regression problem,and solving the classification regression problem has a good solution in the field of machine learning algorithms.This article is inspired by the XGBoost algorithm's brilliant performance in various data mining contests.At the same time,in the study of the existing quantitative investment stock selection model,it is considered that some stock selection feature factors may contain very little information,but it There are also some particularly critical pieces of information.If you use the XGBoost algorithm model directly,you may ignore these factors when performing feature splitting.And there are correlations between the feature factors and the information is also repeated.Considering the appeal problem,this paper adds Principal Component Analysis(PCA)to improve the XGBoost algorithm.The PCA algorithm and XGBoost algorithm are combined to form a new P-XGBoost algorithm model.The new P-XGboost model is used to synthesize feature selection factors.The main information can also reduce the feature factor dimension.At the same time,based on the characteristics of the PXGBoost model itself,it has two particularly good advantages.One is that when the splitting feature factor is selected,the calculation speed is faster,there is no correlation between the samples and the feature factors,and parallel calculation can be realized.Compared with other models,the calculation speed is greatly improved;Regular terms are added to the algorithm to prevent overfitting of the model and make the model more generalizable.Finally,the P-XGBoost model was used as a quantitative investment selection model in the stock market.The P-XGBoost model and the improved XGBoost model were compared in the stock market.The P-XGBoost model was used to forecast the stock price trend.Effective and feasible.