PREDICTING FOOTBALL MATCH OUTCOMES USING LARGE LANGUAGE MODELS: A COMPARATIVE STUDY WITH TRADITIONAL MACHINE LEARNING METHODS

Accurately predicting football match outcomes is valuable for stakeholders such as fans, analysts, sports betting companies, and team strategists. In this study, we explore the potential of large language models (LLMs) for predicting football match results by transforming numerical features into contextual inputs. Key features include historical match results, player ratings, coach ratings, and other relevant conditions, which are processed by the LLM to predict the match winner. We compare the performance of LLM-based predictions with traditional machine learning (ML) models, including random forest and XGBoost. Our findings demonstrate that LLMs achieve comparable accuracy to these conventional ML techniques. Additionally, the LLM offers a significant advantage in that it requires no model training, simplifying implementation and reducing computational costs. This makes LLMs a promising, resource-efficient alternative for football match prediction, presenting new opportunities for AI-driven sports analytics.