Predicting League of Legends Match Outcomes Through Machine Learning Models Using Past Match Player Performance

League of Legends (LoL) is a popular multiplayer online battle arena (MOBA) game with a significant competitive player base. Despite its skill-based matchmaking system, achieving ideal balance remains challenging due to variations in player performance based on chosen champions and roles. This study analyzes over 11,000 ranked matches, using player metrics like Overall Win Rate, Role Win Rate, and Champion Win Rate to train machine learning models: Gradient Boosting, Logistic Regression, and Deep Neural Networks. The results show that all three models achieved an average accuracy of $97.5 \%$, with Champion Win Rate emerging as the most important predictor of match outcomes. These findings provide insights into how past player performance influences match outcomes.