基于波动择时绩效的已实现协方差预测模型比较
发布时间：2018-10-11 18:47
【摘要】：波动择时策略是一种根据资产波动以及相关性构建投资组合的方法,具有较为广泛的应用。鉴于此,提出以波动择时绩效的经济意义指标比较已实现协方差矩阵的预测模型。用高频数据构建股指期货、国债期货和黄金期货的已实现协方差矩阵,利用简单移动平均模型、指数加权移动平均模型和混合数据抽样回归模型对协方差矩阵进行一步向前滚动窗预测,然后在均值-方差框架下根据预测协方差构建动态投资组合,并通过经济效益指标对不同模型的预测进行比较评价。实证结果表明,在股市上升阶段,用简单长期移动平均模型预测已实现协方差矩阵时波动择时策略表现最好;在股市下跌阶段,用简单短期移动平均模型则更优;而用指数加权移动平均和混合数据抽样回归模型时波动择时策略表现则始终居中。
[Abstract]:Volatility timing strategy is a method to construct portfolio based on asset volatility and correlation, which is widely used. In view of this, a prediction model of covariance matrix is proposed to compare the economic significance index of volatility timing performance with that of realized covariance matrix. The realized covariance matrix of stock index futures, treasury bonds futures and gold futures is constructed with high frequency data, and a simple moving average model is used. The exponential weighted moving average model and the mixed data sampling regression model are used to predict the covariance matrix in one step forward rolling window, and then the dynamic portfolio is constructed according to the prediction covariance under the framework of mean-variance. The prediction of different models is compared and evaluated by economic benefit index. The empirical results show that in the rising stage of the stock market, it is better to use the simple long-term moving average model to predict the realized covariance matrix, and to use the simple short-term moving average model in the stage of stock market decline, and to use the simple short-term moving average model to predict the realized covariance matrix. However, the performance of volatility timing strategy with exponential weighted moving average and mixed data sampling regression model is always in the middle.
【作者单位】： 南京大学工程管理学院;
【基金】：国家自然科学基金资助项目(71201075,71671084) 高等学校博士学科点专项科研基金资助项目(20120091120003)
【分类号】：C934
，
本文编号：2264867
[Abstract]:Volatility timing strategy is a method to construct portfolio based on asset volatility and correlation, which is widely used. In view of this, a prediction model of covariance matrix is proposed to compare the economic significance index of volatility timing performance with that of realized covariance matrix. The realized covariance matrix of stock index futures, treasury bonds futures and gold futures is constructed with high frequency data, and a simple moving average model is used. The exponential weighted moving average model and the mixed data sampling regression model are used to predict the covariance matrix in one step forward rolling window, and then the dynamic portfolio is constructed according to the prediction covariance under the framework of mean-variance. The prediction of different models is compared and evaluated by economic benefit index. The empirical results show that in the rising stage of the stock market, it is better to use the simple long-term moving average model to predict the realized covariance matrix, and to use the simple short-term moving average model in the stage of stock market decline, and to use the simple short-term moving average model to predict the realized covariance matrix. However, the performance of volatility timing strategy with exponential weighted moving average and mixed data sampling regression model is always in the middle.
【作者单位】： 南京大学工程管理学院;
【基金】：国家自然科学基金资助项目(71201075,71671084) 高等学校博士学科点专项科研基金资助项目(20120091120003)
【分类号】：C934
，
本文编号：2264867
本文链接：https://www.wllwen.com/guanlilunwen/lindaojc/2264867.html
