基于小波-高阶矩模型的投资组合策略——以国际原油市场为例

Portfolio Strategy Based on Wavelet-High Order Moments model-Take the International Crude Oil Markets as An Research Objects

考虑到投资者异质性特征,将极大重叠离散小波变换方法与高阶矩投资组合框架相结合,提出小波-高阶矩投资组合模型,在此基础上提出频域视角下的高频尺度集成方案和时-频域视角下的全尺度集成方案,并遴选出合适的风险偏好特征改进模型,最后进行稳定性检验。基于国际原油市场数据,样本外检验结果表明:相较于对照组,大部分的小波-高阶矩投资组合策略均取得了更优的投资效果,其中集成部分表现最佳,且高频尺度集成方案侧重于提升收益,而全尺度集成方案侧重于降低波动;通过选择合适偏好高阶矩风险的特征,将会明显改善原始小波-高阶矩投资组合策略,且对两个集成方案改良效果最显著;稳健性检验证实了以上结论。

The existing high-order moments portfolio model doesn’t take into account the heterogeneity of investors,ignoring the value of multiple time-scales,which is difficult to meet the investors’diversified needs.Therefore,the Maximal Overlap Discrete Wavelet Transform method is combined with the highorder moments portfolio framework to propose the Wavelet-High-order moments portfolio model.By constructing high-order moments portfolios after decomposing the time series,this model can meet the diversified needs of investors in different trading cycles.Then,using the idea of“first decompose and then integrate”,a high-frequency scale integration scheme from the frequency domain perspective and a fullscale integration scheme from the time-frequency domain perspective based on the Wavelet-High-order moments portfolio model are proposed.In addition,according to the different risk preference of investors,the appropriate risk features are selected to improve model.Finally,the stability test is carried out.In view of the drastic fluctuation of international crude oil markets,the effect of the portfolio model is tested based on the data of the international crude oil markets.WTI,Brent and Dubai Crude Oil are taken as the research objects.The time spans from October 11th,2006to May 16th,2018,with a total of 2800trading days after deleting non-common trading days.The data of WTI and Brent are obtained from EIA,and the data of Dubai Crude Oil is obtained from Phoenix Financial Network.The results of the out-of-sample test indicate that,compared with the control groups,most of the Wavelet-High-order moments portfolio strategies have achieved better investment results,with the integration part performing best.And the high-frequency scale integration scheme focuses on improving revenue,while the full-scale integration scheme focuses on reducing fluctuation.By selecting appropriate preference characterristics,the original Wavelet-High-order moments portfolio strategy will be significantly improved,and the improvement effect of the two integration schemes is most significant;The robustness test confirms the above conclusions.This study broadens the research scope of high-order moments portfolio theory,and has theoretical value and practical significance for investors in crude oil market to optimize asset allocation,prevent and resolve market risks.