Research On Quantitative Investment Strategy Based On Integrated Algorithm

With the rapid development of China’s economy,investors have much deeper understanding of investment,and the way of investment is more scientific and reasonable.Due to the complexity of stock market and limitation of traditional investment method,quantitative investment has become the hot topic.Considering that the stock data have massive and high dimensional characteristics,this paper would discuss the characteristic selection and data modeling in quantitative investment by means of integrated algorithm,which is a classic algorithm in machine learning.According to the characteristics of the stock market,this paper selects the common index types such as technology,growth and quality as candidate factors for research.The first part builds an Adaboost predictive model based on random forest,correlation and mutual information.First,the effect of a single stock on the large market and candidate factors are combined to form a new feature.And then the characteristic selection is made by using correlation analysis,mutual information analysis and random forest feature importance assessment.Finally,the Adaboost model is used to predict the opening and closing prices of the CSI 500 Index.The empirical results show that the determination coefficients for the opening and closing prices are 0.9744 and 0.9794,respectively,and the fitting degree is much better.The second part builds a timing strategy based on random forest,LightGBM and RSRS.Firstly,the characteristic importance of candidate factors is evaluated using random forests.And then the LightGBM algorithm and resistance support relative strength timing index are taken to establish stock selection strategy based on the selected influence factors above scoring 0.01.The empirical results show that the net value of the strategy is 0.868,an increase of 0.328 over the base net value,and an annualized yield increase 21.21%,which can beat the whole market benchmark.The empirical analysis of this paper effectively verifies the validity of the mathematical model based on the integrated algorithm,which has some reference value for quantitative investment.