新型小波支持向量机在波动率预测中的实证研究

Empirical Research on a Novel Wavelet Support Vector Machine in Volatility Prediction

运用支持向量机进行广义自回归条件异方差(GARCH)模型预测所面临的一个主要问题就是普通核函数难以准确捕捉股指波动率的聚集特征。然而小波函数却具备以任意时间粒度在任意位置刻画任一时间序列的能力。因此,本文基于小波分析与核函数理论,构造了一个满足mercer条件的多尺度小波核来解决这一问题。通过真实股指数据分析,小波支持向量机在波动率预测中的有效性获得了证实。

One of the main problems in generalized autoregressive conditional heteroscedasticity（GARCH） model forecasting is that general kernel functions in support vector machine（SVM） can＇t capture the cluster feature of volatility accurately. While wavelet function yields features that describe of the volatility time series both at various locations and at varying time granularities, so this paper constructs a multidimensional wavelet kernel function and a multi-scale wavelet kernel meeting the mercer condition to address this problem. The validity of wavelet support vector machine （WSVM） in volatility prediction is confirmed through experiments on real-world stock data.