The article introduces a new neural network called VANAR that can predict future states of complex systems and identify cause-and-effect relationships more accurately than traditional methods like VAR. VANAR was tested on chaotic systems and real-world economic data, showing superior forecasting and causality detection compared to other models. While both VANAR and VAR struggled to predict the impact of external shocks on chaotic systems, VANAR proved more robust in handling various types of dynamics.