Strategy And Empirical Analysis Of Improved SVM Multi-factor Stock Selection Model

In recent years,with the prosperity of the Internet and artificial intelligence technology,quantitative investment has gradually come into people's eyes.It uses a variety of mathematical models to solve the problem of how to select stocks and choose the timing.After writing the program,it performs mechanized ordering operations to avoid subjective influences.Compared with traditional investment,quantitative investment is characterized by its refinement and quantification.Taking Buffett,a representative of traditional investment,as an example,the company he founded can obtain a compound return rate of more than 20% per year under the concept of sound management and lasted for 40 years.As a representative of quantitative investment,Simmons,the hedge fund company he created has an average annual compound yield of more than 35%,which is not difficult to see the importance of quantitative investment in the financial market today.This article uses the Shanghai and Shenzhen 300 as a dynamic stock pool,using the data from 2012 to 2014 as the training set of the model,and the data from 2015 to 2017 as the verification set of the model,trying to build a quantitative stock selection model that can outperform the market.The stock selection strategy is mainly divided into two steps.In the first step,50 stocks are selected as the candidate stock pool of the month through the multi-factor scoring model.The candidate factors are all fundamental factors,and the selected stocks can be considered long-term promising.The second step is to select 10 short-term selected stocks from the candidate stock pool through the improved support vector machine model.The position adjustment time is the first trading day of each week.The candidate factors in this part are all short-term technical.index.In addition,in the model construction,this article conducted a horizontal comparison and discussion on the selection of factor weights in the multi-factor model and the parameter setting method in the support vector machine model.In the backtesting of the model,the factor rotation and the rolling windows of the training set were also considered,and added backtesting comparisons in different markets.It established a dynamic quantitative stock selection model.The final improved model achieved an excess return of 15.63% between 2015 and 2017,with an annualized return rate of 19.04% and a Sharpe ratio of 0.71,far exceeding the benchmark performance over the same period.