Research On Quantitative Investment Strategy Based On Artificial Intelligence Technology

In recent years,artificial intelligence has become a research hotspot in academic circles.It has been widely used in image recognition,natural language processing and other fields,and also has a great impact on the field of quantitative investment.As an investment method to obtain stable returns through quantitative statistics,model construction and automated trading,quantitative investment is deeply loved by financial institutions and investors.At the same time,as an important application field of quantitative investment,quantitative investment strategy based on artificial intelligence technology has emerged.How to apply AI to quantitative investment so as to better realize high profit and risk control has also become the focus and difficulty of research.This dissertation constructs a quantitative investment strategy based on artificial intelligence technology,selects the monthly factor data of CSI 300 component stocks from January 2006 to December 2021 as the research object,obtains factor data with the help of Join Quant platform.Factors data include basic factors,emotional factors,growth factors,momentum factors,pershare factors,quality factors,risk factors,style factors and technical factors.At the first,random forest model and factor correlation are used for single-class factor effectiveness test and factor selection,the results are used to construct factor combination.Then,random forest,support vector machine,BP neural network,LSTM model and Attention-LSTM model are used to construct quantitative multi-factor stock selection strategy.The initial capital of quantitative strategy back-test trading is set to 1 million,and the funds are allocated according to equal weight to buy the securities.The stock selection model uses monthly data to predict the future performance of individual stocks,adjusts alternative list of stock monthly,buys the top 10% of stocks,and sells poor stocks in instead.The effectiveness of different AI algorithms in quantitative investment strategies are comprehensively analyzed and compared with the evaluation indicators of strategic return rate,Sharp ratio and maximum retracement.The results of the single-class factor effectiveness test shows that the basic factors and quality factors perform best in the multi-factor stock selection strategy,while the growth factors,momentum factors and risk factors perform indifferently.After factor selection and redundant factor elimination,compared with the performance of five different factors combination,64 factors are determined as the optimal combination,followed by 48 factors combination and 90 factors combination,and the 9 factors combination and 260 factors combination performed generally.Finally the combination is tested in different stock selection models,thus the results fully illustrate the scientific and effectiveness of factor selection using artificial intelligence technology.Compared with the results of different models,the strategy yield of Attention-LSTM model is 1223.10%,followed by random forest,LSTM model,support vector machine and BP neural network.Comparing the model prediction accuracy rate,the Attention-LSTM model is up to 0.69,followed by random forest,LSTM model and support vector machine,and BP neural network is only 0.57.Based on the back-test performance of the quantitative multi-factor stock selection strategy constructed by the five types of models show that without considering the training time and experimental cost,the Attention-LSTM model is the best choice,the second is random forest model.LSTM model and support vector machine can also be selected as reference strategies.