Research on Prediction of Football Match Results Based on XGBoost and LSTM

There is a wide variety of data involved in football matches, including static and dynamic data such as team history and odds data. This diversity leads to limitations in the predictive accuracy of single models for football match outcomes. In order to improve the accuracy of prediction of football match results, this paper uses a combination of XGBoost and LSTM models to predict football match results. By using the collected static data such as player rankings, team rankings, weather conditions, etc., as features, input the XGBoost model to predict the probability of the game result, and then use the predicted outcome probability as the model's prior knowledge, fuse it with the odds change data, and input it into the LSTM model to get a more accurate prediction of the game result. Experimental results show that the proposed method has achieved good results in the prediction of football match results, and provides a useful reference for the research and practice in related fields.