Research On Multi-factor Quantitative Stock Selection Model Based On Machine Learning

With the improvement of computer technology and the rapid development of quantitative investment in developed overseas countries,overseas financial institutions have developed a mature and comprehensive multi-factor quantitative model that can continue to excavate excess returns from the market.In China on the other side of the ocean,due to the late start of the securities market,the immaturity of the market and the inadequate mechanism have led to a lot of arbitrage space for transactions.Domestic institutional investors have learned from the relevant foreign theories and used the quantitative and systematic nature of strict investment Executing trading instructions and other benefits have made a lot of profits,so quantitative investment has also received more and more attention from the Chinese securities industry and has gradually become a hot issue for industry research.With the continuous advancement of China’s financial reform,quantitative investment will inevitably flourish in China’s capital market and show a strong development space.This article is mainly based on the multi-factor stock selection theory under quantitative investment as the background for the multi-factor stock selection model effectiveness research,is based on the rice basket quantitative platform and its database,through machine learning ridge regression algorithm to build a multifactor stock selection model.The sample data selected in this paper are the financial indicators and other factor indicators of the Shanghai and Shenzhen 300 component stocks from 2015 to 2018.The selection of factors takes into account market experience and the degree to which factors affect stock returns.In this paper,IC analysis is conducted on candidate factors,and the factors with higher scores are selected by scoring method.Next,correlation analysis is performed on the selected factors to synthesize the factors with higher correlation to determine effective factors.Finally,effective factors are established with The ridge regression equation of the stock’s next period return yields the final regression stock selection model.In the empirical test,try to introduce timing mechanism and hedging mechanism in order to improve the effectiveness of the model stock selection and the actual reference value.This article introduces the construction and optimization of multifactor models in detail,which has a very positive effect on the current understanding and correct application of multi-factor models.Of course,this article also has a lot of shortcomings,I hope this article can inspire readers to explore the multi-factor stock selection strategy,and jointly participate in the improvement of China’s quantitative investment cause.