In order to predict the winners (and losers) of the 2006 Sundance Film Festival we began by gathering up a ton of data on all the films that had been in the Dramatic and Documentary competitions as far back as 1996.

Specifically, got Film Guides for the last 10 years of the Festival. These Film Guides served as a kind of input. They are written by the Sundance selection committee members all of whom, necessarily, have seen the movies to be screened.

There is a lot of data these reviewers leave as nuggets of information in every film description. Where the film will screen, when it will premier, how many times it will be shown, and who will actually write the review that appears in the Film Guide all give rise to a pattern that helps indicate future success. For each of the more than 360 films in the Dramatic and Documentary Competitions over the last ten years we gathered over 200 data points.

To determine the ultimate success of these films, we then scoured the Internet Movie Database (IMDB.com) and other sources of movie information. We took into account user ratings of films, box office receipts, and other measurable factors to quantify a movie's success. We also took into account any awards given in the Sundance competition itself with special weight given to the Grand Jury Prize and Audience Awards.

With this data we were able to use statistical models in order to train a super-duper computer on the patterns revealing what makes a successful movie. We used John Graham-Cumming's POPFile, a Naive Bayesian classifier, to crunch the data and develop pathways to predict success at future festivals.

By entering the 2006 Film Guide data, and running that data through the statistical pathways established by prior years' festivals, we are able to make predictions on the success of this year's films. Testing the system with known data from previous years, we have established an approximately 81% typical accuracy rate on a year-by-year basis.

Please remember that the predictions are all in fun. You shouldn't make multi-million dollar decisions with a studio's money based on this analysis. On the other hand, if you buy one of the movies that we've predicted to be a winner, and you need a little extra support justifying your decision, our engineers would be happy to attend your acquistion party in order to talk about the incontrovertible proof of box office success the math provides!