Do Your Oscar Predictions Stack Up? Here’s What The Data Says

Predicting the Oscars is tough. We can’t poll Oscar voters, and we don’t know all that much about them. In the absence of direct data, we have to find a way to get inside the voters’ heads to find out who might win.

But whether you’re participating in an Oscar pool or wagering heavily on Irish gambling sites, everyone following the race is trying to figure out who has the edge. The FiveThirtyEight model tries to evaluate the state of the race by figuring out which award shows — like the Directors Guild and the Golden Globes — have historically predicted the winners.

Our model has worked pretty well in the past, but here at FiveThirtyEight, we think it’s always worth trying to find new ways to solve hard problems. Over the past month, we’ve talked to eight amateur modelers who think they’ve found a better way to gauge the state of the Oscar race.

Let’s revisit them each briefly:

• First, we looked at models that try to mine the Internet to figure out who could win an Oscar. Burak Tekin’s model uses Google News; Paul Singman’s pulls from tweets. Both of these models get more and more accurate the closer we get to the big day — Feb. 28.
• Next, we looked at models that take in either a little bit or a whole lot of data to figure out the Oscars. Brian Goegan built a model that looks at earlier award shows and additional nominations to figure out a winner. Zach Wissner-Gross and Randi Goldman boiled their data down to box office dollars and Rotten Tomatoes scores to find their leaders.
• Third, we checked out models that try to get inside the heads of people who think like the Academy. James England asks people to vote on films and performances they’ve seen in head-to-head matchups. Nigel Henry and his crew at Solution by Simulation analyze the MovieLens data set to find people who gave high ratings to films that won Oscars in the past and to see which films they liked this time around.
• Lastly, we looked at two models that analyze what’s been written about movies to pick winners. Allison Walker analyzes film reviews for words that have historically been used to describe Oscar winners. Gary Angel and his team at Ernst & Young analyze cultural publications and film reviews to find which Oscar nominee best fits with the current Hollywood worldview.

Oscar voting began Friday and will continue through Feb. 23. With just over a week to go before the ceremony, we checked back in with each of the modelers for their current picks.

The best picture race has been a doozy: “Mad Max: Fury Road” got off to a strong start in our model, dueling with “Spotlight” in the early part of January. Then “Spotlight” began to pull ahead, only to fall behind “The Big Short” after the highly predictive Producers Guild awards. Once the Screen Actors Guild awards had been announced, however, the two films were neck and neck — until “The Revenant” pulled off huge wins at the Directors Guild and BAFTAs, the awards from the British Academy. Our model has “The Revenant” with a decisive lead going into the Academy Awards, but “The Big Short” and “Spotlight” can’t be counted out.

Here’s what each of the models says as of Tuesday. Not all the modelers were able to convert their picks into probabilities — I mean, we don’t either, so I hardly blame them — but for those who felt confident enough to put a number to it, we have the current probability:

Although there’s some love for “Mad Max: Fury Road,” “The Big Short,” “Room” and “Brooklyn,” it looks like our two front-runners are “The Revenant” and “Spotlight.”

In our model, this category comes down to the winner of the Directors Guild award. Before Alejandro G. Iñárritu won it for “The Revenant,” George Miller, who directed “Mad Max: Fury Road,” had scooped up most of the big wins. Among the seven modelers who were able to apply their methodology to this category, all but one have one of those two directors ahead.

There appears to be near consensus here that Leonardo DiCaprio (“The Revenant”) will win. In 1,000 years, aliens from a star 999.5 light-years away will arrive on Earth with an intent to destroy it if Leo doesn’t get the shiny statue. And we will deserve it.

The lone dissent: The Ernst & Young team finds that Bryan Cranston (“Trumbo”) is ahead in the race. (Angel reports that their model for best actor is still a bit of a work in progress.)

There’s a surprising amount of variation in this category! Our model decisively has Brie Larson (“Room”) in the lead. But among the guest modelers, every nominee appears in at least the top two, and four are the leaders in at least one model.

This category could prove a solid test for methodological soundness. The models that back Larson display a strong degree of confidence, though: Goegan’s even says it’s a virtual lock. Perhaps a consensus will emerge, but honestly, it’s way more fun this way.

Six of the models were able to adapt to make a prediction in the supporting categories and they are split.

Our model has Kate Winslet (“Steve Jobs”) in second place after she got a bump from the BAFTAs, but it’s not going to be enough to take the lead away from Alicia Vikander (“The Danish Girl”). The models that look at Internet buzz seem to have picked up on this recent surge and are picking Winslet. The ones that look at the language of reviews, the head-to-head preferences of people who saw the films, and award wins have Vikander on top. This could be a fun one.

I’m going to come right out and say it — this category is a catastrophe. The front-runner in our model, Sylvester Stallone (“Creed”), hasn’t won anything in a month; the winner of the Screen Actors Guild award, Idris Elba (“Beasts of No Nation”), wasn’t nominated for an Oscar; and the early favorite, Mark Rylance (“Bridge of Spies”), is only on the board because of a last-minute win at the BAFTAs. I have very low confidence in our state-of-the-race model’s ability to provide legitimate insight into the eventual winner: It’s a coin toss at best and a dice roll at worst.

Can these folks help us out?

Nope. Everyone’s all over the map here. This should be another fantastically fun category to watch. I hope Mark Ruffalo (“Spotlight”) wins because I want to be able to say “Academy Award winner and also The Hulk, Mark Ruffalo” on first reference moving forward.

The Rest

Several of the modelers went out of their way to submit predictions outside of the top six categories.

Goegan sent in results for all the feature-length categories. We’ll hit the lot of them next week, but he has “The Big Short” and “Spotlight” winning best adapted and original screenplay, respectively, “Son of Saul” taking best foreign film, “The Revenant” winning the two sound awards and cinematography, and “Mad Max: Fury Road” winning the editing, costumes and makeup prizes.

Goegan, Singman, Henry’s team and Zach and Randi all said “Inside Out” is the odds-on favorite to win best animated feature. Singman has “Inside Out” also winning best original screenplay and “The Big Short” winning best adapted screenplay. Henry’s team has “Mustang” winning best foreign film and “Star Wars: The Force Awakens” winning best visual effects and best score.

We’ll talk to all of these folks again next week for final predictions going into the Oscars. Until then, check out our ongoing coverage and our interactive tracking the state of the Oscar race.

More Predict The Oscars:

Can You Read Between The Lines To Pick The Oscar Winners?

The 2016 Oscars Race

How Much Do We Need To Know To Predict The Oscars?

Can The Internet Predict The Oscars?

FiveThirtyEight’s Guide To Predicting The Oscars

Can You Fake The Academy To Predict The Oscars?