Cornell Ann S. Bowers College of Computing and Information Science – “For years, the financial press has helped inform investors of all stripes. Cornell researchers have discovered it can also inform the algorithm behind a new financial predicting model.  In their paper, “News-Based Sparse Machine Learning Models for Adaptive Asset Pricing,” published in Data Science in Science in April, the researchers draw from interdisciplinary fields such as machine learning, natural language processing (NLP) and finance to build a new, interpretable machine-learning framework that captures stock- and industry-specific information and predicts financial returns with greater accuracy than traditional models. “One of the knocks on machine learning is it’s not interpretable,” said Martin Wells, the Charles A. Alexander Professor of Statistical Sciences in the Cornell Ann. S Bowers College of Computing and Information Science and the paper’s senior author. “Often when researchers use big models such as these, they may not know what the outputs mean or what is underlying the model. This research leverages text data from the news to build interpretable machine-learning models where you can see the important features explicitly.”