Using SVM with Financial Statement Analysis for Prediction of Stocks

At present, there are many technical analyses for prediction in stock market. However, the technical indices are fluctuated with the quantity of stock exchanges. The financial indices are more reliable, nonvolatile and valid compared with the technical indices. In this paper, we propose an original and universal method by using SVM with financial statement analysis for prediction of stocks. We applied the SVM to construct the prediction model and select Gaussian radial basis function (RBF) as the kernel function. The experimental results show our method not only improve the accuracy rate, but also meet the different stockholders’ expectations. INTRODUCTION Support vector machine (SVM) is a useful technique for data classification (Burges, 1998), regression (Smola et al, 1998) and prediction (Müller et al, 1997). Previously there has been a lot of study using artificial neural network (ANN) in these areas, especially in the field of prediction. However, in the stock market, because the data often has enormous noises and complex dimensionality, the ANN method has some limitations (Kim, 2003). Recently, SVM has been successfully used in the field of prediction. SVM can treat higher dimensional data better even with a relative low amount of training set. Further more, it can present a good ability of generalization for complex model (Thissen, 2003). Some applications by SVM to predict the stock market have been issued, but the degree of accuracy rate and the acceptability of certain prediction are measured by the predictors’ deviation from their own experiences or the ineffective data (Huang, 2005). In the field of prediction for stock market, the most important thing is to improve the prediction accuracy rate (Huang et al, 2006. Chen et al, 2006). However, little study has justified the suitability of stock market prediction by SVM. In this paper, we propose an original and universal method by using SVM with financial statement analysis for prediction of stocks. Commonly there are many technical analyses for prediction in stock market. But these technical indices such as RSI, BIAS, etc. appear to fluctuate with the quantity of stock exchanges. Compared with the technical indices, the financial indices from the financial statement are much more reliable, nonvolatile and valid. The goal of this paper is to improve the accuracy rate of prediction and to meet different kinds of stockholders’ expectations. This paper is organized as follows. In section 2, we will briefly explain the theory of SVM and some concepts from finance and accounting. In section 3, the methodology is given. In section 4, the experiment and the experimental result is shown. In section 5, we will present the conclusion and some suggestions. Communications of the IIMA 63 2007 Volume 7 Issue 4 Using SVM with Financial Statement Analysis for Prediction of Stocks Han & Chen