A satisficing approach to factor portfolio construction

Multifactor models and the construction of factor-matching portfolios are by now pervasive in investment management. In textbook discussions, the construction of factor-tracking portfolios is presented as a simple exercise solved using well-known optimization methods such as linear programming. While mathematically correct, the standard approach does not address two practical considerations. First, when building portfolios, investment managers typically consider matching the target loadings of some factors as more important than matching the target loadings of other factors. Second, investment managers generally seek to achieve their investment objectives in the most efficient way possible, by minimizing turnover and/or transaction costs. In short, portfolio construction generally requires prioritizing and reconciling multiple objectives. To address the shortcomings of the standard factor-matching framework, we show how a multi-objective optimization methodology known as 'goal programming' can be effectively employed to build portfolios that successfully balance competing investment aims.