Factoring Qualitative Views into Portfolio Insights via Machine Learning

The article covers a brief piece of proprietary research exploring a relatively new machine learning approach to enhance portfolio optimisation problems: the TIC (Theory-Implied Correlation) matrix approach. This new approach allows to factor qualitative information into the portfolio optimization process to enhance robustness.Although the example presented in this article has been focused on minimising tracking error from the standpoint of an ETF provider concerned about synthetically replicating a liquid index, the TIC approach could also be integrated by active investment managers seeking a different objective such as limiting the amount of risk of their mandates, or increasing their reward per unit of risk