A1007
Title: Maximally machine-learnable portfolios
Authors: Maximilian Goebel - Bocconi University (Italy)
Philippe Goulet Coulombe - Université du Québec à Montréal (Canada) [presenting]
Abstract: When it comes to stock returns, any form of predictability can bolster risk-adjusted profitability. A collaborative machine learning algorithm is developed that optimizes portfolio weights so that the resulting synthetic security is maximally predictable. Precisely, MACE is introduced, a multivariate extension of alternating conditional expectations that achieves the aforementioned goal by wielding a random forest on one side of the equation, and a constrained ridge regression on the other. There are two key improvements with respect to Lo and MacKinlay's original maximally predictable portfolio approach. First, it accommodates any (nonlinear) forecasting algorithm and predictor set. Second, it handles large portfolios. Exercises are conducted at the daily and monthly frequency and significant increases are reported in predictability and profitability using very little conditioning information. Interestingly, predictability is found in bad as well as good times, and MACE successfully navigates the debacle of 2022.
