Application Research Of Multi-factor Stock Selection Based On AdaBoost Algorithm

Machine learning is a newly developed computer technology,and more and more scholars are beginning to explore how to combine this technology with the investment in the stock market in order to achieve high yields.The AdaBoost algorithm is a method of machine learning.The algorithm first constructs a weak classifier that judges the correct rate slightly higher than the random guess.Then,by repeatedly learning the data samples,the weak classifier is finally trained to become the strong classifier with extremely low error rate.The purpose of this paper is to construct a multi-factor stock selection model based on AdaBoost algorithm by combining machine learning algorithm(AdaBoost algorithm)and multi-factor quantitative stock selection method to achieve excess return that higher than the benchmark income.In order to construct this stock selection model,the factors needed to construct the model must first be screened out.In this paper,244 factors in the uqer quantization factor pool are selected as the candidate factors,and the factors must be screened and tested for independence.Finally,10 factors are selected as the ?independent effective factor? of the model.The model uses these 10 factors as weak classifiers to repeatedly learn the stock's rate of return and identify the stock according to the stock yield.Those stocks ranked in the top of 30% of the stock return rate are identified as 1 and those ranked in the bottom of 30% are identified as-1.The AdaBoost algorithm first assigns the same weight to all stocks,then updates the weight of the stocks in the portfolio according to the classification of the weak classifier.If the classification is correct,the weight is reduced,and if the classification is error,the weight is increased.The weak classifier is repeatedly trained through the AdaBoost machine learning algorithm,and finally combines the weak classifiers to a strong classifier with high classification accuracy.The multi-factor stock selection model based on AdaBoost algorithm performs well in the backtesting process,with an annualized rate of return of 25.5%,which is 22.4% higher than the benchmark yield of HS300 shares.