Financial time series forecasting based on wavelet kernel support vector machine

Financial time series forecasting is a hot research topic in the field of finance and it is of great significance to the study of finance market. Since the nonlinear characteristics of financial time series and the shortcomings of traditional forecasting methods, a new support vector machine (SVM) based on wavelet kernel function and its construct algorithms are proposed. Furthermore, the wavelet kernel functions have been proved to satisfy the admissible condition. The new SVM models are applied to forecast the Nasdaq composite index to test their forecasting performance. Compared with polynomial kernel SVM and Gaussian kernel SVM, experimental results show the wavelet kernel SVMs can increase the prediction accuracy, enhancing prediction model generalization performance.