Designing time-series models with hypernetworks and adversarial portfolios

This article describes the methods that achieved fourth and sixth place in the forecasting and investment challenges, respectively, of the M6 competition, ultimately securing first place in the overall duathlon ranking. In the forecasting challenge, we tested a novel meta-learning model that utilizes hypernetworks to design a parametric model tailored to a specific family of forecasting tasks. This approach allowed us to leverage similarities observed across individual forecasting tasks (i.e., assets) while also acknowledging potential heterogeneity in their data generating processes. The model’s training can be directly performed with backpropagation, eliminating the need to rely on higher-order derivatives, and is equivalent to a simultaneous search over the space of parametric functions and their optimal parameter values. The proposed model’s capabilities extend beyond M6, demonstrating superiority over state-of-the-art meta-learning methods in the sinusoidal regression task and outperforming conventional parametric models on time series from the M4 forecasting competition. In the investment challenge, we adjusted portfolio weights to induce greater or smaller correlation between our submission and that of other participants, depending on the current ranking, aiming to maximize the probability of achieving a good rank. While this portfolio strategy can increase the probability of securing a favorable rank, it paradoxically exhibits negative expected returns.