Predicting financial time series is challenging due to volatility and nonlinear dynamics. Traditional methods often forecast single series, overlooking critical inter-stock relationships. To overcome this, we adapt the Attention-Based Spatio-Temporal Graph Convolutional Network (ASTGCN)—originally developed for traffic prediction—to forecast stock prices across multiple companies. Using spatio-temporal graph techniques, we rigorously evaluate ASTGCN against advanced baselines, including the Dynamic Spatial-Temporal Aware Graph Neural Network (DSTAGNN) and the Adaptive Spatio-Temporal Graph Neural Network (ASTGNN). We also model inter-stock relationships among nine companies using adjacency matrices. The results demonstrate that ASTGCN achieves high predictive accuracy, establishing it as a reliable and scalable solution for multi-company stock prediction.