Better risk and performance estimates with factor-model Monte Carlo

Yindeng Jiang and R. Douglas Martin

Abstract

A common problem in asset and portfolio risk and performance analysis is that the manager has such a short history of asset returns that risk and performance measure estimates are quite unreliable. But the manager has available long histories of many risk factors and can use a subset of them to construct a high-R-squared risk-factor model for the asset returns. We introduce a simple method of simulating from such a factor model that yields considerably more accurate risk and performance measures, and show that it is important to use a statistically justified method of choosing the risk factors. The resulting factor-model Monte Carlo method works well by adequately reflecting the nonnormality of the factor and asset returns, and by borrowing strength from the correlation between the risk factors and the asset returns.