Posted by Rob Knies

David Rothschild

It’s mid-February, and the nominations for the 86th Academy Awards have been public for a month. Fans are scrambling to watch all the nominated films they can find. Film buffs debate the talents of Blanchett and Bullock until closing time. Excited partygoers and party throwers are planning feverishly.

David Rothschild, too, has his own Oscars preparations in the works. He’s not watching the movies, though. Instead, he’s focusing on the data associated with them.

You might recall that Rothschild, an economist from Microsoft Research New York City, made his initial foray into predicting the winners of the Oscars in 2013—and that his forecasts proved accurate in 19 of the 24 Academy Awards categories.

Now, he is attempting to improve upon that already gaudy record in the days leading to March 2, when members of the film community gather at Hollywood’s Dolby Theatre for the opening of the envelopes from the Academy of Motion Picture Arts and Sciences.

“This is only the second year I have been doing this,” he says, “and I was shocked how confident I was last year, but they were nothing like this year.

“First, 2013’s confidence was later in the cycle, after the major awards shows had passed. We did see a lot of movement in several major categories last year. Second, that confidence did not extend to the big categories as aggressively as it has this year. Of the biggest eight categories, my predictions are already averaging 84 percent, and I expect to get more confident as the event approaches.”

Rothschild employs data from prediction markets such as Betfair, the Hollywood Stock Exchange, and Intrade, along with other user-generated data.

The goal is to make cost-effective predictions, not only so they can scale to all 24 Oscars categories—most people who even attempt to make such predictions limit themselves to just the most ballyhooed awards—but also to learn to make useful forecasts in a variety of domains.

He also is offering predictions that evolve in real time, from the announcement of the Academy Awards nominations until the awards themselves are revealed.

“I learned a lot from 2013 that will help me make more accurate predictions in 2014,” he divulges. “My colleagues and I worked through statistical models, unique polling data, and several forms of prediction-market data for 2013. We saw the pros and cons of the various methods and have incorporated new methodology for 2014.

“We also collected a lot more data in 2013 than we did in previous years, helping us to refine our 2014 estimates.”

So, what do his data indicate? You can see all the latest probabilities at PredictWise—and yes, he doesn’t just pick winners, he assigns them percentages of likely success—but to pique your interest, here are a few:

  • Best Director: Alfonso Cuarón (Gravity), 98 percent.
  • Best Leading Actor: Matthew McConaughey (Dallas Buyers Club), 90 percent.
  • Best Supporting Actress: Lupita Nyong'o (12 Years a Slave), 62 percent.

Those of you itching to learn whom Rothschild is predicting to win best costume design or best sound mixing will just have to visit PredictWise. (I’m predicting that 83 percent of you will.) But Rothschild notes that the only major category that remains a barnburner is Best Original Screenplay, in which he calculates that Her leads American Hustle, 50 percent to 44 percent.

It’s scientific, but it’s fun, too, harnessing the kinds of topics that people discuss during social gatherings.

“I love the fact that my work is enjoyable to people at cocktail parties!” Rothschild laughs. “But I concentrate on domains where the data are available and people are willing to participate and provide data.

“That being said, we are very excited to extend some projects soon into economics and business questions. I look forward to demonstrating the value of this prediction work in meaningful questions that will help people efficiently allocate resources.”

That, too, figures to be an accurate forecast.