中文摘要：

提出了运用非参数方法SVR与改进的期权定价方法结合的期权价格预测模型．首先利用股票价格收益率的偏度和峰度对传统的期权定价方法计算出期权的价格进行修正．然后，通过引入非参数方法SVR对其结果进行拟合来减小传统参数模型的误差，并建立SVR滑动窗口预测模型．由于传统的方法不能有效的把握实际期权价格的运动趋势和非线性的特点，所以在第一阶段的预测后，在第二阶段引入SVR来解决其非线性，进而减小误差．最后，利用我国长虹CWBl权证以及随机10只认购权证日价格数据进行实证检验．结果表明：在预测精度方面，非参数方法要优于传统的参数方法，而改进后的期权定价方法比传统的方法更符合实际情况．

英文摘要：

This paper proposed an improved option price forecasting model using the nonparametric method support vector regressions （SVR） combined with improved option pricing models. First, we verified the conditional option price with skewness and kurtosis of return of the corresponding stock. Second, we employed SVR to fit the results, decreasing the errors of the parametric methods, and set up the forecast model of SVR with sliding windows. Since the conventional methods mimic the trends of movement of the real option prices, with these methods in the first stage, SVR could concentrate its power in nonlinear curve approximation to further reduce the forecast errors in the second stage. Finally, extensive experiments with data of Warrant Changhong CWB1 and ten more random call warrants of China＇s warrant market demonstrated that the forecast accuracy of nonparametric method was better than that of parametric method and the improved option pricing methods were more practical than the conditional ones.