Research On Machine Learning Multi-factor Stock Investment Strategy In New Energy Vehicle Industry

New energy vehicle industry,as a major emerging industry in the world,is in the midst of rapid development and great change.Since 2009,China’s new energy vehicle industry has accumulated a certain first-mover advantage and scale advantage.At present,China has the largest number of new energy vehicle brands and automobile manufacturers in the world.Since the rise of the technology industry is closely related to the development of the capital market,the development expectation of the global capital market for the new energy vehicle industry is also reflected in the securities market.The overall investment style of China’s A-share market is gradually becoming rational,value-oriented and growth-oriented.In addition,with the rapid development of Internet information technology,quantitative investment schemes based on artificial intelligence have great development potential in the future financial field.Development environment based on the above,this paper using machine learning science,build more factor model,in order for data selection and pretreatment,model selection and back-test,strategy optimization,in-depth analysis of the empirical work,such as the final a-share market of China will be applied to construct A set of new energy automobile plate dynamic stock investment strategy,and presented the better results.In this paper,22 alternative factors were selected from various authoritative research reports,and 13 effective factors were finally screened out to constitute the factor pool in this paper through IC and IR value test,correlation between factors,characteristic importance analysis,etc.Secondly,missing values,outliers and standardization of effective factors were preprocessed to improve the accuracy and efficiency of model training.Then,three models of different subdivisions were selected for training and backtest,and the Stockranker ranking model with the best overall performance was selected to proceed to the next optimization step.Furthermore,in order to further improve the rigor and return rate of the strategy,we try to add delisted stock and ST stock processing and the trading logic of the market risk control mechanism.The results show that the return rate and the maximum retracement of the strategy are significantly optimized.Finally,through in-depth analysis of the accuracy,benefits and risk degree of the above strategy,the feasibility and authenticity of the strategy is verified from different perspectives.The empirical results show that the dynamic investment strategy of China A-share new energy vehicle stocks constructed in this paper presents the final result of A total return of260.43%,an annual return of 30.38%,A maximum retraction of 24.03% and A Sharpe ratio of 1.11 between 2016 and 2020.On the whole,this strategy has gained certain excess returns.The total returns over the past five years are far higher than the benchmark returns of the CSI300 market,and the maximum pullback is also relatively reasonable.The increased strategy optimization mechanism does play a certain positive role in improving the accuracy and stability of stock selection,effectively improving the return of strategy and reducing the risk.