## Statistically Modeling Oscar

An Oregon scientist develops an algorithm to forecast the winners of this year’s Academy Awards.

Every year around this time, the media seems to be singularly devoted to coverage and speculation regarding the Oscars: Will this year’s host be funny? What will our favorite stars be wearing? What picture will have a big night and take home the most golden statues? Which veteran actor will be upset by a young up-and-comer?

It’s not quite the Super Bowl, but bookies around the world, are currently taking bets on all the major categories, which raises the question: Who should we put our money on?

Obviously, unless you counted the votes, that answer is—No. But Iain Pardoe, a statistician in the department of decision sciences at the University of Oregon, thinks Brokeback Mountain is pretty much a sure bet for best picture—a 91.1% lock, to be exact.

Evaluating various factors, such as previous Oscars won by directors or actors, how many total nominations a particular movie receives and the results of other award ceremonies like the Golden Globes, Pardoe has created an algorithm to predict who will win in Oscar’s four major categories: best picture, best director, best actor and best actress.

“I tried to figure out which factors were the most important in being associated with the actual outcome of the Oscar itself,” Pardoe said. After collecting data from award ceremonies dating back to 1928, he “fed it into a statistical model and turned the crank and tried to figure out what worked the best.”

In Pardoe’s model, Brokeback Mountain holds a significant statistical advantage at the top of the best picture heap because of its wins at the Golden Globe and Producer’s Guild award ceremonies. But the statistician is quick to point out that these same wins didn’t help The Aviator last year, which was upset at the Academy Awards by Million Dollar Baby. That was the only prediction Pardoe didn’t nail on the head last year.

While, Pardoe is using the rigor of math to formulate his predictions, he realizes choosing a winner isn’t just a numbers game—computer algorithms can’t account for the Academy members’ personal preferences and biases.

Since beginning his research last year, Pardoe has retroactively applied his statistical algorithm to past Academy Awards and said the model has performed quite well. In the last 30 years the model correctly predicted the best director winner 93% of the time.

“Lots of people have their own pet theory about who is going to win the Oscar,” said Pardoe. “What I tried to do was look at all the information more objectively and see what the data actually says when you throw everything into the mix.”

This year Pardoe is putting his money on Ang Lee winning best director for his work on Brokeback Mountain. That’s despite this being the sixth year when Steven Spielberg has been nominated (Munich), which inflates his chance at winning. But, Lee’s victory at the Director’s Guild Awards gives him a 90% chance of taking home a little golden man, according to Pardoe.

In the category of best actor, Pardoe’s model favors Capote’s Philip Seymour Hoffman, who has a 94.8% chance of walking away with the Oscar.

The prediction Pardoe is least certain about is for best actress. His algorithm chose Reese Witherspoon for her role in Walk the Line.

Witherspoon won the Golden Globe for best actress in a musical/comedy this year, but Felicty Huffman also won a Golden Globe for best actress in a drama for her role in Transamerica. Both actresses are up for the Oscar, and while Pardoe’s model says Witherspoon has a reasonably hefty 75% chance of winning, he admits that this is his model’s least confident prediction.

While only Pardoe will have to hold his breath in anticipation four times during Sunday’s Oscar ceremony, he said that if he has more time in the future, he is going to put together a model to predict best supporting actor and actress, also. In the meantime, time remains to change your bet on Sunday’s winners—no one need know math was on your side instead of a gut feeling.

Originally published March 2, 2006

Tags

## Now on SEEDMAGAZINE.COM

• ### Ideas

#### I Tried Almost Everything Else

John Rinn, snowboarder, skateboarder, and “genomic origamist,” on why we should dumpster-dive in our genomes and the inspiration of a middle-distance runner.

• ### Ideas

#### Going, Going, Gone

The second most common element in the universe is increasingly rare on Earth—except, for now, in America.

• ### Ideas

#### Earth-like Planets Aren’t Rare

Renowned planetary scientist James Kasting on the odds of finding another Earth-like planet and the power of science fiction.