A multi-factor analysis model of quantitative investment based on GA and SVM

In the field of securities investment, how to choose the right stock is always an essential subject to investigate. In order to successfully identify the fluctuation trend of stock-price, this paper proposes a multi-factor analysis model that is based on the genetic algorithm (GA) and support vector machine (SVM). The price of stocks is affected by many factors; however, the extent of influence is not the same. If simply removing factors of small influence, we would lose certain details of data. In this paper, we use GA to automatically weight features of various factors so as to highlight factors having great influence, as well as to reduce factors having little effect. Meanwhile details of the stock data will be retained. The SVM algorithm can be further used to classify these factors and finally to identify the stock price fluctuation trend. The empirical analysis shows that the aforementioned quantitative stock-selection model is feasible and effective, which prevails over the Decision Tree algorithm, the k-Nearest Neighbor algorithm and the Naive Bayesian algorithm.