Can Big Data predict the Best Picture Oscar winner?
With many struggling to provide insights from Big Data Bryan Melmed puts his money where his mouth is by using it to predict the winner of the Best Picture Oscar.
Can big data insights predict the winner of Best Picture at Sunday’s Oscars? We think it can and we are betting the farm on 12 Years a Slave. And this is how we figured it out.
Melmed
Firstly, we know that the winners are chosen by less than 6,000 votes, cast by a group the Los Angeles Times discovered are older (86 per cent are over 50), white (94 per cent), and male (77 per cent).
By virtue of our data about people who work in the entertainment industry and fit this wealthier, LA based demographic, we know they have a love of sports cars, high-end clothing and exotic destinations. Using this as a starting point, we created lookalike models using 15,290 of the topics on our data platform and, based on their online behaviour, compared the interests of users with a persistent and lasting interest in the movies nominated for Best Picture and those of the Academy panel demographic.
Of course this also involves a degree of human supposition and extrapolation. But that is the essence of data; ‘Big’ or small. Regardless of the volume, granularity and quality of available data, you need people who can apply accumulated expert knowledge to unlock the insights that data, even highly contextualised data, suggests. Without this human element the promise of Big Data will never be realised.
Four Non-Contenders
We were quickly able to scrub Captain Phillips, Her, Philomena, and Nebraska from the race.
Captain Phillips is the least popular title searched for in the Hollywood area. The previous film choices of Captain Phillips fans also make it a dud. Its fans also liked Warm Bodies (a romantic zombie comedy), Identity Thief (a crime comedy), and The Last Stand (an action film with Arnold Schwarzenegger and Johnny Knoxville) making it too accessible to be considered.
Digital marketers dream date-movie Her, is also a non-starter with a fan base that is far more into science fiction and technology themes (almost 20 times more likely) than the average quinquagenarian of the Academy. They are also just too rich to vote for Philomena. Its fans are the least affluent of the nominees, being 2.8 times more likely to earn less than $50,000 a year.
Nebraska sits in a different corner as the true art house favourite, with a young urban audience that is decidedly aspirational – in fact, we see that roughly half are spending well more than they earn.
Two Near-Misses
Both The Wolf of Wall Street and Dallas Buyers Club are serious contenders, but won’t be making it to the final round for Best Picture. You could say that one is too selfish and the other not selfish enough.
The Wolf of Wall Street has been criticised for glorifying greed and even exploiting women. Hollywood doesn’t necessarily avoid controversy, but fans of Wall Street play to stereotype, which will likely make it distasteful to the Academy. It’s not just that they are incredibly wealthy — 9.6 times more likely to earn more than $250,000 a year, 6.8 times more likely to be investment bankers, and 4.6 times more likely to buy an imported Italian suit. It’s also that many seem stuck in a pre-recession mindset. We see high interest in “established” celebrities like Heidi Klum (10.7 times), Megan Fox (8.4 times), and Britney Spears (8.3 times). These fans wear Abercrombie and Fitch (8.2 times) or Lacoste (4.2 times), drive a Hummer (3.3 times), and a few are still using Blackberry phones (2.5 times). In other words, the problem isn’t that the film condones excess, but that there’s something quite unfashionable about it.
At the other extreme, the Dallas Buyers Club fans are undeniably a caring bunch. They are the most likely to be expecting a child (11 times), to be dog owners (8.3 times), to be vegetarian (7.3 times), to be elementary school teachers (6.5 times), and even to watch “Dr. Phil” (4.8 times) in Portland (4.1 times). (They’re also twice as likely to buy Volvos and air purifiers.) Possibly perpetuating the stereotype of a heartless, rapacious Hollywood, but you’d be unlikely to find this audience less than 350 miles north of LA. There’s a psychological element here as well. In general, people prefer movies they can feel good about watching (such as 12 Years A Slave) to watching a movie about people doing good.
Three Challengers
Removing the outliers is easy. What about the three remaining contenders? American Hustle, Gravity, and 12 Years a Slave enjoy broad appeal, with none of the baggage of the other films.
A simplistic demographic approach would suggest a win by American Hustle. As viewers get older or wealthier, they increasingly prefer the film. Caucasians are most likely to be fans as well.
Dig just a little deeper, and a different pattern emerges.
Consider geography. Not only are denizens of Los Angeles more likely to be fans of 12 Years, but so are moviegoers in New York, San Francisco, Atlanta, Miami, Washington D.C., and Seattle. American Hustle fans are found in cities on a different cultural axis (if you will), including Chicago, Philadelphia, Minneapolis, Denver, Phoenix, and Kansas City. The only metro area preferring Gravity is Austin (a city whose unofficial slogan is “Keep Austin Weird” so often plays against type).
Or we can explore with the film category itself. The most highly correlated movie titles are in the table below. Which set is most likely to be in an Academy member’s library?
Still, this is only scratching the surface. With any one approach, we leave behind everything else we know. Let’s finally consider the interaction of all the variables we’ve identified: older white men living in Los Angeles who (broadly speaking) has a love of sports cars, high-end clothing, and exotic travel destinations. We can also guess they are socially well-established, with urban and coastal cultural preferences. What of the users on our network who fit that profile?
As you can see, our prediction is that 12 Years a Slave takes the Academy Award for Best Picture.
And if it doesn’t, we’ll be back with a better model next year.
Bryan Melmed is director of insights at Exponential Interactive base din New York and will be speaking at ad:Tech in Sydney.
Nice ‘big data’ PR stunt, but I know a 66 year old who can pick a winning horse in a field of 24 every time without “indexing” the participants! Suggestion: run your ‘model’ against the results of the last five years. If the model picks the winner in a majority of these years, then you’re onto something. Big data doesn’t need a future result to validate its accuracy. Oh, and there’s a reasonable correlation between the winning film and adspend in Variety.
User ID not verified.
Congrats.
User ID not verified.
Well done.
Also, doesn’t the methodology prove films popular amongst rich, white, old, male west-coast Americans win Academy Awards?
Not sure why the endorsement of the planet’s least popular demographic would count for much.
User ID not verified.
Replace ‘big data’ with ‘probability’.
User ID not verified.