OscarLytics: The Math Alone Predicted 80 Percent of the Winners
Although statistical analysis didn't point to "12 Years a Slave's" best picture winner, it correctly called 16 of 20 categories at this year's Academy Awards.
The Oscars are a delightful celebration of the arts – from storytelling to acting to design to music – but as it turns out, the Oscars are a fantastic venue for celebrating math as well.
For the third year in a row, my mathematical formula for predicting the Oscars broke the 75 percent mark, this time going 16-for-20, good for an 80 percent success rate. In many of the night’s most exciting races, statistics proved to have the answer.
For best supporting actor, my algorithm – which relies only on historical data from other awards shows, critic scores, guild awards, and previous Oscars – correctly chose Jared Leto from Dallas Buyers Club (48 percent) over Barkhad Abdi from Captain Phillips (38 percent).
For best original screenplay, the math gave stronger weight to the Writers Guild win for Her (45 percent) than to the BAFTA win for American Hustle (38 percent), another accurate pick.
Playing the numbers was an equally smart strategy for close calls in best foreign-language film (The Great Beauty – 40 percent), best documentary (20 Feet from Stardom – 49 percent), best production design (The Great Gatsby – 29 percent), and best costume design (The Great Gatsby – 44 percent).
The statistics also correctly rewarded many categories that were fairly easy to pick with appropriately high percentages: Alfonso Cuaron for best director (69 percent), Matthew McConaughey for best actor (71 percent), Cate Blanchett for best actress (76 percent), Frozen for best animated feature (a record 92 percent), and Gravity for best sound editing (65 percent), best sound mixing (60 percent), best cinematography (70 percent), and best visual effects (85 percent).
Fortunately, the numbers were not perfect. I say "fortunately," because in my mind it would take all of the suspense out of a thrilling awards show if data alone could predict all the winners. Two of the misses were fairly obvious well in advance: The math said Gravity for best picture because of its Directors Guild win – usually an excellent predictor – but in a year so nicely set up for a relatively rare picture/director split, as I explained in my pre-Oscars predictions column, it makes sense that the Directors Guild would only predict best director, not best picture. Similarly, in the adapted screenplay category, the math knocked 12 Years a Slave down to second place because of its ineligibility at the Writers Guild (which I do count differently from a WGA loss or lack of a nomination) and its BAFTA loss.
Impressively, only two additional categories saw mathematical upsets. The first was in best film editing, where Gravity had a lower percentage than Captain Phillips because of losses at the Eddies (run by the American Cinema Editors) and the BAFTAs, but the Academy voters were clearly smitten with Gravity in just about all technical fields (deservedly so, I would say). The second was best supporting actress. The math predicted Jennifer Lawrence, mainly because no one in history had ever lost best supporting actress after winning both the BAFTA and the Golden Globe. Even if the anecdotal evidence rather clearly favored Lupita Nyong’o, I unfortunately can't take such evidence into account numerically.
All of this math talk might seem a bit dry, but for me, it makes the Oscars come to life. Because hidden beneath all of the 1s and 0s I keep stored in Excel spreadsheets (17,820 data points, to be precise) is the very history of film. It's the story of winners and losers, of landslides and upsets, and of one grand film after another. Behind each of those 1s and 0s is a story of a young woman who ventured from Kenya to the Yale Drama School and the Dolby Theatre stage. It's a story of a beloved fairy tale that journeyed from a Danish storybook to Walt Disney's imagination to a computer-animated masterpiece. It's a story of unbelievably talented singers who were finally pushed 20 feet forward into the spotlight they so deserved, joining thousands of (I will refrain from saying "countless") entertainers over the past 86 years and hopefully many more to come.
I can’t wait to join you next year to see what stories the movies and math will share with us in 2014.