Bing Remains Horsepower in Predictive Modeling

Bing Scores 20 Out of 24 With Oscar Predictions

In the world of predictive modeling, Microsoft’s Bing just added another feather to its cap.

In the recent past, Bing and the Microsoft research team have correctly predicted the Super Bowl, Grammy awards and the World Cup winners. Now, Microsoft’s prediction engine predicted 20 out of 24 winners of the Academy awards. The four predictions that Bing missed all came in second place.

Bing correctly picked the winners of the best supporting actor, best supporting actress, best actor, best actress, best director, best picture and many others. The four categories it missed in were best original screenplay, best animated feature, best original score and best film editing; beating many psychic readings.

“Perfectly calibrated 20 of 24 (w/ 4 2nd place winners),” tweeted David Rothschild, the head of Microsoft’s New York team and developer of the prediction model, after the awards show was over. The prediction model actually performed better last year when it picked 21 out of 24 winners.

“Prediction markets follow a select group of people who have high levels of information on what voters will do and are willing to wager real-money on the outcomes,” said Rothschild in an interview with TechNet. “And, prediction market-based forecasts have been incredibly accurate,” which accounts for the successful predictions of previous years’ Oscar winners.

The Bing results were much more accurate than Las Vegas-compiled odds for the different categories. Bookmakers in Vegas predicted that Boyhood would win Best Picture and Michael Keaton would win Best Actor. Of the six predictions that Wynn casino made, they were accurate with just four. In those categories, Bing scored a perfect six.

Having a successful track record in predictive modeling is important if Bing intends to replace Google as the industry standard for predictive technology. Rothschild noted that, ahead of the Oscars, most of the awards are “relatively unpredictable.”

“The main goal is to show people that Bing algorithms and data itself is a pretty powerful force in terms of what we can do,” said Walter Sun, manager of Bing’s Core Ranking Team, in a blog post. “Being able to just parse out this information to predict a winner for a voting contest, or the order of a draft that’s coming up, or the likely outcome of the World Cup – things are an interesting way to show users that Bing has a lot of horsepower beyond just providing good search results.”