Multi-Factor Stock Selection Model Of SVM

With the development of the theories of Mathematical Finance, Behavioral Finance, etc. and the boom of Computer Technology, Quantitative Investment has been gradually promoted theoretically and methodologically and its idea has become easier and easier to be accepted by domestic investors. Particularly, under the boost of China's financial reform, Quantitative Investment shows infinite charm and development space. Multi-factor Model, an important part of Quantitative investment, is the basis for the building of many quantitative models as well as a current hot issue in the field of China's Quantitative Investment. Keeping searching more efficient multi-factor stock selection models is significant to the development of Quantitative Investment theory and the operation of brokers and investment funds.This thesis studies on the effective factor combination of China's A share market and than building of Support Vector Machine(SVM) Multi-factor Stock Selection Model. Firstly, through organizing the existing multi-factor stock selection models at home and abroad, determine the thesis' research perspective and idea; based on the research idea, elaborate the relevant theories and concepts of multi-factor stock selection model and SVM and structure a theoretical framework for SVM multi-factor stock selection model; then test the model Empirically –basing on the thought of Stepwise Regression, select effective factor combination of China's A share market and via dividing the market style, discover two extra factors; and then take effective factor combination as input feature set, apply SVM Classification Algorithm and build SVM Multi-factor Stock Selection Model. At last, conclude this thesis' research efforts and results and put forward my views on future research directions and ideas.This thesis applies SVM method to build Multi-factor Stock Selection Model and does an empirical research on the stock of A share market from August 2005 to October 2014. The results show that the effective factors of China's A share market include net profit growth rate of parent company, enterprise value multiple TTM, operating margin, turnover rate, Price-to-sales ratio TTM and return on equity of parent company; that distinguishing different market style features can discover some hidden effective factors; that using SVM method, the Multi-factor Stock Selection Model is steady which improves the validity of the model and provides ideas for the progress of multi-factor stock selection model.