A multi-criteria approach to evolve sparse neural architectures for stock market forecasting

The development of machine learning based models to predict the movement of a financial market has been a challenging problem due to the low signal-to-noise ratio under the effect of an \textit{efficient} market. Although, researchers have developed neural network based predictive models to address this issue, the selection of an appropriate neural architecture is seldom addressed. This study, therefore, proposes a new framework to evolve \textit{efficacious} yet \textit{parsimonious} neural architectures for the movement prediction of stock market indices using technical indicators as inputs. The proposed approach formulates the neural architecture search as a multi-criteria optimization problem to balance the efficacy with the complexity of architectures. In addition, the implications of different underlying trading tendencies, which may be present in the pre-COVID and peri-COVID time periods, are investigated. An $\epsilon-$constraint framework is proposed as a remedy to extract remaining concordant information from the possibly partially conflicting pre-COVID data. Further, a new search paradigm, Two-Dimensional Swarms (2DS) is proposed for the multi-criteria neural architecture search, which explicitly integrates sparsity as an additional search dimension in particle swarms. A detailed comparative evaluation of the proposed approach is carried out by considering genetic algorithm and several combinations of empirical neural design rules with a filter-based feature selection method (mRMR) as baseline approaches. The results of this study convincingly demonstrate that the proposed approach can evolve comparatively efficient and parsimonious networks.