Predicting the outcome of races in Formula 1 is a challenging task, because race outcome prediction depends not only on driver skills but also on vehicle performance and relative strategies among teams and conditions for races. This work introduces a novel approach to hybrid predictive modeling that integrates conventional machine learning methods with advanced graph neural networks to predict the outcome of races in Formula 1. Our model uses a comprehensive dataset employing techniques like hyperparameter tuning and SMOTE to enhance performance. The output of the optimized model, a GridSearch-based Random Forest with SMOTE, was then used as input for a GNN to further refine predictions. The GNN’s ability to model relational and temporal data enabled a deeper understanding of driver and vehicle dynamics, leading to more nuanced and accurate race outcome predictions.This hybrid approach leverages the Random Forest’s ability to handle high-dimensional, non-linear data with the GNN’s superior capability to capture intricate relational and temporal dynamics, such as driver interactions and race conditions. It doesn’t only contribute to sports analytics but also opens doors to wider avenues for research and applications of predictive models in more complex, dynamic scenarios.