Data-based identification of short term predictors for stock market trends using heterogeneous model ensembles

We here show the application of heterogeneous ensemble modeling for training short term predictors of trends in stock markets. A sliding window approach is used; model ensembles are iteratively learned and tested on subsequent data points. The goal is to predict trends (positive, neutral, or negative stock changes) for the next day, the next week, and the next month. Several machine learning approaches implemented in HeuristicLab and WEKA have been applied; the models produced using these methods have been combined to heterogeneous model ensembles. We calculate the final estimation for each sample via majority voting, and the relative ratio of a sample's majority vote is used for calculating the confidence in the final estimation; we use a confidence threshold that specifies the minimum confidence level that has to be reached. We show results of empirical tests performed using data of the Spanish stock market recorded from 2003 to 2013.