Statistical models have been increasingly utilised in recent years in football, providing valuable insights into player performance, team tactics and overall match outcomes. Expected Goals (xG), a statistical measure, has been in the spotlight for development as it provides meaningful insights to analysts and coaches on a team’s attacking efficiency and identifies areas for improvement. In this paper, the StatsBomb API open data was utilised and the imbalance in the dependent variable classes was addressed using Synthetic Minority Oversampling Technique for Nominal and Continuous characteristics (SMOTENC). Models such as Logistic Regression and Light Gradient Boosting machine (LightGBM) were employed to further improve the xG model. Through the evaluation using accuracy metrics such as the Brier Score, ROC AUC, and McFadden’s pseudo R-squared, it was observed that LightGBM outperformed other models, achieving a Brier Score of 0.0684.