To create a classifier, we first gathered statistics from baseball-reference.com on every game played in the last 15 years. Initially, we created features from the average of the players’ batting statistics (average, OBP, slugging, OPS) for each team and the starting pitcher’s ERA for each team. We tried many different classifiers on this dataset using WEKA to determine the top few performers for 10-fold cross-validation accuracy, which turned out to be AdaBoost, J48 decision trees, simple logistic regression, and voted perceptrons. With that we narrowed our focus on these four methods, and we hoped to create new attributes more specific than just the average of player statistics for a game that improved testing accuracy.