Si-Xing Liu

Enhancing TimeXer with TCN and differential attention for multivariate stock forecasting

DAT-TimeXer is proposed, a structure-aware adaptation of TimeXer for closing-price forecasting that achieves the lowest mean forecasting errors among the compared models in the one-step evaluations and maintains lower errors at horizons of 1, 3, 5, and 10 in the evaluated multi-step tasks.