Reducing Your Reliance on Risk Models: Another Look at Active Share
9 Pages Posted: 22 Oct 2014
Date Written: October 20, 2014
Risk models play a key role in quantitative equity management. While risk models are generally good predictors of ex-post portfolio volatility we are painfully aware that, at times, these models are subject to significant misspecification. For this reason we cannot view risk model predictions as point estimates but must view them instead as confidence intervals, within which the true measure of risk should lie. In this paper we demonstrate that for any given risk model misspecification, the width of a portfolio’s confidence interval is a positive function of its active share. In other words, the higher a portfolio’s active share the less confidence we have in its risk measurements. Monte Carlo simulation shows that these confidence intervals grow nonlinearly with active share.
Keywords: Active Share, Risk Model, Smart Beta, Monte Carlo, Factors
JEL Classification: G11
Suggested Citation: Suggested Citation