Regularized Mimicking Portfolios

We propose new approaches to constructing mimicking portfolios for non-tradable shocks from a large set of base assets. The key element of our procedure is the imposition of regularization constraints on portfolio strategies that help mitigate the overfitting problem caused by a large number of statistical moments that determine optimal portfolio weights. We empirically explore the out-of-sample performance of mimicking portfolios for nine macroeconomic and uncertainty shocks obtained by applying the proposed techniques. In all cases, our mimicking portfolios have less extreme weights than those produced by standard methods without sacrificing the portfolio performance. When shocks can be mimicked by stock returns, the performance of our portfolios is superior to the performance of their unregularized counterparts