Advertisement

Oscars 2015: Big Data number crunchers try their hand at calling the awards

Everyone knows what those briefcases held last year. The trick is to guess the winners before Sunday.
(Al Seib / Los Angeles Times)
Share via

As almost anyone who’s ever joined an office pool can tell you, Oscar predictions are famously squishy affairs, likely to flummox even a sophisticated film fan.

But in recent years, Big Data has attempted to turn art into science. This year such efforts are more prolific than ever — and more relevant, with the battle for best picture (presumed by pundits to be between “Boyhood” and “Birdman”) considered one of the more wide open in recent memory.

Whether these analyses are accurate, though, remains to be seen. What is clear is that their number and methods are growing, as enterprising quantitative types now build demographic models, comb betting markets and even break down nominees’ narrative elements, all in a bid for Nostradamus status.

Advertisement

FULL COVERAGE: Oscars 2015

Given that the Academy of Motion Picture Arts and Sciences is an anonymous body of about 6,600 — and not, say, a market with decades of history or an election with dozens of finely honed polls — there’s no reason to think Big Data will do better than the insider pundits (or dart-throwers) at predicting the Academy Awards. In fact, members of the data community have acknowledged this, notably in a FiveThirtyEight post last year titled “Why It’s Hard to Predict Oscar Winners” that concluded that “for now, a market-beating Oscar prediction model is probably out of the picture.”

Still, that hasn’t stopped many outlets, including FiveThirtyEight itself, from trying.

Here are some of the more notable efforts, including their methodology and their sure-fire, numbers-tested winners for this year.

Advertisement

Exponential

The ad network doesn’t look at previous winners but instead examines the strains of prototypical Oscar voters, drawing on 32,000 Web users from the millions in its own network. After building a model that it believes reflects the Oscar voting body, it then seeks to find the nominated pictures that enjoy the highest correlation between its fan base and the academy’s demographics.

For instance, “The Theory of Everything” is favored by those who want Elizabeth Warren to run for president — “Northeastern liberals [who] are huge fans of NPR, work for nonprofits or educational institutions, and favor electric hybrids like the Toyota Prius.” That’s a group that Exponential believes doesn’t, on the main, describe Oscar voters, who it says are “Hillary Clinton voters, liberal but not left-wing, educated but not intellectual, and with a West Coast affinity for strong personalities.”

Advertisement

Prior accuracy: The company said “12 Years a Slave” and “Argo” were the leading contenders for best picture in the previous two years. It turned out to be true, though given what pundits were saying, these were not market-beating predictions.

2015 predictions: “Birdman” and “The Imitation Game,” Exponential says, are the two leading contenders for best picture. If the latter pulled an upset it would prove the service may be on to something. “Birdman”? Not so much.

PHOTOS: Oscars 2015 top nominees | Presenters | Nominee reactions | Awkward moments

PredictWise

This website, run by David M. Rothschild, an economist with Microsoft Research, aggregates and analyzes data from prediction markets like Britain’s Betfair and the Hollywood Stock Exchange. It also factors in other data, such as previous awards.

Prior accuracy: Last year, coming into the awards PredictWise gave “12 Years a Slave” an 85.8% chance of winning best picture and Alfonso Cuarón (“Gravity”) a 98.6% chance of winning the director Oscar. Both predictions were correct, though again, many pundits also called those races correctly.

Advertisement

2015 predictions: The service has this year’s race as much less of a runaway. Still, it believes “Birdman” is twice as likely to win best picture as “Boyhood” (66.8% to 32%). For director, the site predicts “Birdman” helmer Alejandro G. Iñárritu has a somewhat better chance than “Boyhood” director Richard Linklater (59%-40.7%).

FiveThirtyEight

Nate Silver’s number-crunching website has based its predictions on the results of 16 awards ceremonies leading up to the Oscars, according to a weighted system based on their track records and memberships.

Prior accuracy: In 2013, the site offered a strong probability that “Argo” would win best picture — along with many pundits — and saw its prediction come true Oscar night. In the race for best director, FiveThirtyEight considered it a near-tossup between Steven Spielberg (“Lincoln”) and Ang Lee (“Life of Pi”). The site gave Spielberg a slight edge, but Lee won.

2015 predictions: In a best picture contest described as “pretty tight,” FiveThirtyEight has “Birdman” ahead of “Boyhood,” with the former garnering 3.38 of the site’s points compared to 2.47 for “Boyhood.” For director, the site gives Linklater a slight lead on Iñárritu, with 2.02 polling points to 1.72.

OSCARS 2015: Complete list | Ballot | Cheat Sheet | Timeline | Oscars 2015 bingo

Advertisement

MovieGraph

Run by the data company Senzari, the outlet looks not at other wins or quantifiable data but the more amorphous plot and tonal elements of recent winners. Among the factors it studies is whether a film contains “tragedy” (appearing in more than half the winners in the last decade) as well as “morality.” It also, for reasons less clear, factors in box office, using more dollars as one indicator of an Oscar.

Prior accuracy: N/A

2015 predictions: “American Sniper” has the highest score among these key elements and is thus forecast as best picture this year.

Psychics

Because there’s little empirical evidence that statistical modeling offers a better chance at accuracy than your neighborhood palm reader, let’s look at what they say. There’s the Psychic Source’s “Arthur,” who used his “intuitive gifts” to make predictions. Meanwhile, there’s psychic-to-the-stars Thomas John, who tells The Times his methodology is based on a process called “automatic writing” in which, as he puts it, “I connect, close my eyes, get in a meditative state and start to write things.”

Prior accuracy: Arthur doesn’t appear to have been feeling it last year, but John has made predictions in the past, such as in 2013, when he prophesied that “Lincoln” would win best picture. (It didn’t.)

Advertisement

2015 predictions: John has “Birdman” for best picture and “Boyhood’s” Linklater for director, while Arthur has a “Boyhood” sweep of the top two awards. Arthur may also be reading Oscar blogs as much as the stars: He predicts a runaway win for Julianne Moore in lead actress and “get[s] a close tie” between Eddie Redmayne and Michael Keaton for lead actor.

Times staff writer Amy Kaufman contributed to this report.

Advertisement