TY - JOUR T1 - Factor Allocation as Reverse Attribution JF - The Journal of Portfolio Management DO - 10.3905/jpm.2021.1.225 SP - jpm.2021.1.225 AU - Joseph Simonian Y1 - 2021/02/22 UR - https://pm-research.com/content/early/2021/02/22/jpm.2021.1.225.abstract N2 - Multifactor models and the construction of factor portfolios are by now pervasive in investment management. Active factor allocation can be challenging, however, given the difficulty in generating accurate point estimates for ex ante factor alphas. To overcome this challenge, the present article describes a technique that gives managers the ability to generate expected factor alphas from factor volatilities. An important component of the framework is the way it characterizes factor allocation as reverse attribution. Starting with a portfolio alpha target and risk budget, the methodology proceeds to derive factor weights by finding the benchmark-relative factor loadings that satisfy the alpha and risk objectives. Furthermore, because effective factor allocation is forward looking, the framework described in the article also allows managers to adjust factor volatility values in a manner that is consistent with their subjective beliefs regarding a factor’s likely marginal alpha contribution. The framework presented thus serves as a quantitatively grounded yet practical framework for building actively managed factor portfolios.TOPICS: Factor-based models, portfolio construction, risk management, performance measurementKey Findings▪ Performance attribution can be applied in reverse to inform forward-looking factor allocation.▪ It is possible to derive expected factor alphas from factor volatilities and a target portfolio alpha value and risk budget.▪ A simple, yet statistically driven, approach to calibrating ex ante factor volatilities can be implemented by coupling a manager’s subjective views with an assumed distribution for factor variances. ER -