Kaggle, a site which hosts data science competitions, is hosting a new competition to predict the result of March Madness. What makes this slightly different than your standard office pool is that you predict the result of every possible match-up among the 68 teams by building a model that somehow incorporates historical data. The competition provides some standard data for past seasons and encourages competitors to supplement it with their own data. I’m interested in the results of this, but I doubt I have time to build a model of my own this semester.