440 – 440 - SLDS CSpeed 8
Optimal Financial Portfolio Using Graphical Lasso Under Unstable Environment
Ekaterina Seregina
University of California, Riverside
Unstable environments raise challenges for constructing a financial portfolio. In such scenarios, it is unrealistic to assume constant portfolio weights, whereas estimating weights using only post-break observations omits the information prior to the break point. This paper visualizes stock returns as a network of interacting entities and generalizes network inference in the presence of structural breaks. We estimate time-varying portfolio weights using pre- and post-break data when the stock returns are driven by common factors. Using the example of a strong structural break caused by the first wave of COVID-19 pandemic, we demonstrate that combining pre- and post-break observations for estimating portfolio weights improves portfolio return and Sharpe Ratio compared to constant weights and weights that use only post-break observations.