Customized Risk Analysis through Dynamic Factor Definitions
提出一种新方法，让投资者能动态定义自己的风险因子，并自动将现有投资组合分析映射到新因子上，无需重新估计底层模型，适用于全球多资产组合和增强风险管理。
Linear factor models are the standard framework in the financial industry to quantify and explain sources of portfolio risk. However, most of the commercially available risk models are based on a large number of granular factors, which are predefined and generally updated at low frequency. These can be too granular for global multi-asset-class portfolios, become misaligned with the investment purposes of the model users, or be slow to respond to changing market dynamics or emerging risks. This article proposes a new methodological framework that addresses these challenges. It allows investors to dynamically define their own risk factors in terms of existing model factors or of assets covered by them and to automatically map all their existing portfolio analytics onto their new factors without the difficulty of reestimating the underlying model. The article illustrates the benefits of the framework for a total portfolio approach and for enhanced risk management.