Many researches have been done for stock market prediction. While there are many free and premium data sources available today, and there many new machine learning algorithms have been proposed, most of these researches do not focus on utilizing these data sources and algorithms. This paper proposes a stock market prediction service framework that allows users to choose different data sources and machine learning techniques. In particular, while most existing prediction approaches are based on neural networks, support vector machines or Naive Bayes, we illustrate the flexibility of our framework by including metric learning based methods to predict stock movement.