Abstract: Kernel function in Support Vector Machine（SVM） has a great influence on the quality of model.Currently，in financial time series forecasting，Radial Basis Function（RBF） kernel is primary kernel function，next is polynomial kernel function.However，Every kernel has its advantages and disadvantages.Combining two or more kernels is one of efficient method for improving the ability of learning and generalization.In this paper，RBF and polynomial kernel have been combined to forecast the financial time series.The experiment result shows that the strategy of mixing the kernel function can obtain better performance in financial time series forecasting.

Key words: Support Vector Machine（SVM）, financial time series, mixture kernels

摘要： 核函数是支持向量机（SVM）的重要部分，它直接影响到SVM的各项性能。当前SVM在金融时间序列分析中，基本上采用高斯径向核函数（RBF），其次才是多项式核函数。然而，每种核函数都有它的优势和不足，整合两个或多个核函数对于学习能力和泛化能力的提高是一个有效的途径。采用高斯径向核函数与多项式核函数的混合核函数运用于金融时间序列预测中，且与其单个核函数的支持向量机的实验结果进行了比较。结果表明，混合核函数具有更好的性能。

关键词: 支持向量机, 金融时间序列, 混合核函数