利用非参数回归因子模型的方法估计资产回报率分布并进行投资组合优化的实证研究

发布时间：2018-01-16 10:41

本文关键词：利用非参数回归因子模型的方法估计资产回报率分布并进行投资组合优化的实证研究　出处：《复旦大学》2013年硕士论文　论文类型：学位论文

【摘要】：摘要：Markowitz的均值方差模型提供了一个很好的投资决策的思维框架。但是如果直接使用历史价格数据估计资产回报率的期望和协方差矩阵,并做出控制风险的约束下追求最大收益的最优投资组合决策,在A股市场的实证结果并不理想。改进投资策略可以从两个方面入手：第一,引入新的度量风险的指标；第二,提高对未来回报率的预测能力。本文对这两种思路的实际效果作了实证研究,在第一种思路中,用VaR与CVaR代替标准差,度量投资组合的风险。在第二种思路中,我们建立了一个在现有的信息下估计未来回报率分布的预测模型,用预测模型得到的分布作为投资组合优化模型中的参数。实证结果表明,运用股票的预测模型得到的资产回报率分布进行投资组合决策,可以极大地提升投资表现。而仅仅改进风险度量的指标对于投资组合表现改善不大。
[Abstract]:The mean-variance model proposed by President Markowitz provides a good thinking framework for investment decision, but if the historical price data is used directly to estimate the expectation and covariance matrix of return on assets. And under the constraint of risk control, the optimal portfolio decision to pursue the maximum return is not satisfactory in A-share market. The improvement of investment strategy can be started from two aspects: first. The introduction of new risk measurement indicators; Second, improve the ability to predict the future rate of return. This paper makes an empirical study on the actual effect of these two ideas. In the first way, the standard deviation is replaced by VaR and CVaR. In the second way, we establish a prediction model to estimate the distribution of future return under the existing information. The distribution obtained by the forecasting model is used as the parameter in the portfolio optimization model. The empirical results show that the portfolio decision is made by using the distribution of the return on assets obtained from the forecasting model of stocks. It can greatly improve the performance of investment. But the index of improving risk measurement is not much better for portfolio performance.
【学位授予单位】：复旦大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F224;F830.91

【相似文献】