Working Paper: Improved Stock Price Forecasting Algorithm based on Feature-weighed Support Vector Regression by using Grey Correlation Degree

WORKING PAPER Improved Stock Price Forecasting Algorithm based on Feature-weighed Support Vector Regression by using Grey Correlation Degree Quanxi Wang ABSTRACT With the widespread engineering applications ranging from artificial intelligence and big data decision-making, originally a lot of tedious financial data processing, processing and analysis have become more and more convenient and effective. This paper aims to improve the accuracy of stock price forecasting. It improves the support vector machine regression algorithm by using grey correlation analysis (GCA) and improves the accuracy of stock prediction. This article first divides the factors affecting the stock price movement into behavioral factors and technical factors. The behavioral factors mainly include weather indicators and emotional indicators. The technical factors mainly include the daily closing data and the HS 300 Index, and then measure relation through the method of grey correlation analysis. The relationship between the stock price and its impact factors during the trading day, and this relationship is transformed into the characteristic weight of each impact factor. The weight of the impact factors of all trading days is weighted by the feature weight, and finally the support vector regression (SVR) is used. The forecast of the revised stock trading data was compared based on the forecast results of technical indicators (MSE, MAE, SCC, and DS) and unmodified transaction data, and it was found that the forecast results were significantly improved.

Keywords: support vector regression (SVR), grey correlation analysis(GCA), behavioral finance, stock forecast

1. Introduction With the application of computer science, data processing and data analysis of high-volume financial data have become easier and easier. People prefer to depend on computer to deal with financial problems rather than traditional artificial statistics, especial in stock price forecasting, one of the prevalent parts that many researchers focus on. The discuss with regard to the predictability of stock price in the actual market never ceases, since stock market is a complex nonlinear dynamic system and the estimation and extrapolation of classical nonlinear function values are far from being able to adapt to the complexity of the stock market. Despite its complexity, many machine learning methods stand out to show their applications for stock price forecasting due to their exceptional nonlinear adaptability. Baba N and Kozaki M (1992) use a hybrid algorithm, combined the modified BP (back propagation) method with the random optimization method, to train the parameters in the neural network and develop the algorithm for forecasting stock prices in the Japanese market. Xing Chen (2001) and Weidong Meng and Taihua Yan present a method for stock market modeling and forecasting via fuzzy neural network based on T-S model. Yiwen Yang (2001) and Guizhong Liu and Zongping Zhang provide a method for predicting chaotic data with combining embedding theory and artificial neural networks and apply this method for stock market prediction. Tay F E H and Cao L (2007) propose a modified version of support vector machines to model non-stationary financial time series.
   One of core problems of stock price forecasting is how to choose an effective quantitative financial technique. Support Vector Machine (SVM) is different from the traditional machine learning methods and has exceptional nonlinear adaptability. Its optimization goal is to minimize the confidence range value, and its optimization constraint is the training error. Besides that, The solution problem of SVM is finally transformed into the solution of quadratic programming problem, so the only global optimal solution can be obtained. We usually call SVM Regression SVR when SVM is applied for the regression problems. However, classical SVR method (c-SVR) has some limitations when applying for stock price forecasting since the prediction results are greatly affected by the input feature vector. Therefore, the feature weights of each feature vector should be considered in predicting the model establishment, and certain feature selection should be performed. Grey Correlation Analysis (GCA) is a method for measuring similarity among the factors based on nearness to the models which considering similarity and nearness respectively. By using this method, the feature weight of the feature vector can be determined. Interestingly, feature-weighted SVR (FWSVR) is not a brand-new model. In fact, it has already been examined by James N. K. Liu and Yanxing Hu (2012), developed a feature-weighted support vector machine regression algorithm for the China Shenzhen A-share market. However, until very recently FWSVR has not been considered as a mainstream SVR caused the limited test samples and untested robustness. Another core problems of stock price forecasting is h

…(Full text truncated)…

This content is AI-processed based on ArXiv data.