Oscar Picks: How to Beat Your Film-Geek Friends

You don’t even need to have seen the movies. Let data science and the wisdom of the markets be your guide


I rarely watch mainstream movies.

I don’t know any of Hollywood’s movers and shakers.

Yet I’m confident I can beat the pundits in predicting this year’s Oscars.

I’ll even put some numbers on it: Cate Blanchett has a 99 percent chance of winning best leading actress for her portrayal of an imploding socialite in Blue Jasmine, and Matthew McConaughey a 92 percent chance of bagging best actor for Dallas Buyers Club. I’m also pretty sure 12 Years a Slave will walk away with best picture—I put its chances at 87 percent.

Of the big six Oscars, there’s only one that gives me pause for thought: Lupita Nyong’o looks a powerful contender for best supporting actress in 12 Years, but I wouldn’t call her a lock-in. I give her a 59 percent chance of winning.

Why should you follow my advice, given that Gravity is the only one of the nominees for best picture that I’ve seen?

In a word: data. I’ve been looking at forecasts made by modern-day soothsayers—the practitioners of predictive analytics—and I’m basing my picks on the method that performs best.


Predicting the future is big business. With enough information on patterns of purchasing, for example, data scientists can spot impending life events merely from what’s in your shopping basket. That’s how retailers like Target can reportedly know a woman is pregnant even before she has told her family. You might also remember Nate Silver’s success in predicting Barack Obama’s canter to victory in 2012, and how he embarrassed the political pundits who had told us it would be a close-run thing.

Predicting the Oscars is harder than calling a presidential election because there is no poll taken beforehand of a sample of the people who vote—the 6,000 or so members of the Academy of Motion Picture Arts and Sciences. But there are other data to go on, including box office receipts, ratings from critics, and movies’ release dates. (Conventional wisdom says that producers hold Oscar contenders until late in the annual cycle, so that they are fresh in the memories of Academy members.)

Silver has tried his hand at predicting the Oscars. Twice, he’s scored only four for six in the top categories, and has blamed the limitations of the data for his failure to do better—he once lamented his inability to quantify the “jackassedness” of Mickey Rourke, who was Silver’s favorite for best actor in 2009 for his performance in The Wrestler, but was probably sunk by a history of disparaging remarks about the Hollywood establishment.

I’d be more cautious about my Oscar picks if this sort of statistical modeling was all we had to go on. But there’s another way of predicting future events. It depends not on what’s happened in the past, but on data generated in real time by the wisdom of the crowd.

To see how this works, head to the Hollywood Stock Exchange, one of several websites where people trade in futures markets tied to real-world events. The “Hollywood dollars” at stake there bring kudos rather than real cash. But the process is fun enough to motivate movie buffs to think hard about whether to buy or sell shares in Oscar nominees. As a result, the price of a nominee’s shares reflects the crowd’s best assessment of their chances. Just five days before the Oscars ceremony, 12 Years was trading at more than $14 in the market for best picture, with Gravity a distant second, priced at less than $5.

Why do I think prediction markets provide the best guide to Oscar picks? In a new study, David Rothschild, an economist with Microsoft Research in New York City, has compared the performance of different methods in predicting last year’s winners. He found that picks based on a real-money prediction market called Betfair outperformed the media pundits, and comfortably beat statistical models based on data like box office receipts and release dates.

Those statistical models do yield interesting tidbits of information—perhaps some small-talk for your Oscars party. The Farsite Group, a predictive analytics firm based in Columbus, Ohio, is basing its Oscar picks in part on a statistical model that heavily favors prior awards in the same season, and in part on bookmakers’ odds. It is forecasting a fairly close call for best picture, giving 12 Years a Slave a 55 percent chance of winning, and Gravity 38 percent.

Rothschild agrees that awards handed out earlier in the season do have some predictive power, but says that almost everything else is down in the statistical noise. Even so, you’ll probably be glad to know that audience ratings on Rotten Tomatoes are a better guide to Oscar success than critics’ choices.

All in all, though, Rothschild says that complex statistical modeling adds little to the power of a prediction market. In the past he employed a mixed approach when forecasting the Academy Awards, but this year his Oscar forecasts are derived entirely from Betfair, converted into percentage chances of winning using a tried-and-tested formula.


Many of Rothschild’s predictions, listed on his PredictWise website, are so confident that they make Nate Silver’s eve-of-poll forecast that Barack Obama had a 90.9 percent chance of winning the 2012 election look timid. Rothschild admits to being a little spooked by these numbers. “I would love to temper these predictions, because they scare me,” he told me. “But that’s what the data says.”

On Sunday March 2, all will be revealed. Having stuck my neck out, I’ll just have to fall back on the old cliché if Rothschild’s forecasts prove less than wise: Prediction is very difficult, especially if it’s about the future.


Here, as a handy cheat-sheet for your Oscars party, are Rothschild’s predictions as they stood when I posted this article:

Picture: 12 Years a Slave — 87.4 percent chance of winning

Director: Alfonso Cuarón, Gravity — 98.2 percent

Leading actor: Matthew McConaughey, Dallas Buyers Club — 91.9 percent

Leading actress: Cate Blanchett, Blue Jasmine — 98.6 per cent

Supporting actor: Jared Leto, Dallas Buyers Club — 97.2 percent

Supporting actress: Lupita Nyong’o, 12 years a Slave — 59.1 percent

Adapted screenplay: 12 years a Slave — 92.9 percent

Original screenplay: Her — 54.8 percent

Animated feature: Frozen — 99.1 percent

Animated short: Get a Horse! — 93.1 percent

Foreign language film: The Great Beauty — 86.0 percent

Documentary feature: The Act of Killing — 58.1 percent

Documentary short: The Lady in Number 6 — 91.4 percent

Live action short: The Voorman Problem — 61.7 percent

Original song: Let It Go, Frozen — 89.2 percent

Original score: Gravity — 91.6 percent

Cinematography: Gravity — 98.9 percent

Costume design: The Great Gatsby — 86.2 percent

Film editing: Gravity — 77.3 percent

Makeup/hairstyling: Dallas Buyers Club — 98.0 percent

Production design: The Great Gatsby — 71.2 percent

Sound editing: Gravity — 96.6 percent

Sound mixing: Gravity — 95.2 percent

Visual effects: Gravity — 99.8 percent

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.