One of the big hopefuls at the 2011 Oscars was Moneyball, a surprisingly entertaining film about how the Oakland A’s baseball team used mathematics to pick a winning side.
As a lover of baseball, maths and movies, Ben Zauzmer was eagerly watching the Academy Awards that year.
But the Harvard University freshman mathematics student was more than just a passive observer of the ceremony. He had built his own Moneyball-style system for predicting Oscar winners, and this was his test run.
“I always loved reading about mathematical baseball predictions [and] mathematical political predictions … [and] I assumed that somebody out there had done mathematical Oscar predictions,” Zauzmer said.
“I went to the internet, I started Googling and I found nothing.”
Figuring he had stumbled on an untouched field ripe for statistical analysis, Zauzmer spent a month building a dataset, trawling through previous Oscar winners, other awards shows, even tracking down old press releases from movie launches.
And the winner is…
When Hollywood’s night of nights for the films of 2011 came around (on February 26, 2012), Zauzmer hosted an Oscars party for his friends, and unveiled his mathematical predictions.
“That first year went very well,” he said.
“[In] 15 of the 20 categories, the mathematical favourite went on to win, including all eight of the most prominent categories [best picture, the four acting awards, best director, and the two screenplay awards].”
In some categories, his maths-based predictions even bested the film experts, who had tipped Viola Davis (The Help) to take home the best actress Oscar ahead of eventual winner Meryl Streep (The Iron Lady).
Zauzmer felt he was on to something.
Since then, his mathematical Oscar predictions have been published in the Hollywood Reporter each year.
His strike rate is 77 per cent over the past eight years, with his best score being 20 out of 21 in 2018.
“It goes up and down every year — some years there are a lot of upsets,” Zauzmer said.
Overall, he said his mathematical predictions were on par with the film experts.
“I think that’s because there are both advantages and disadvantages to using only numbers and no hunches like I do,” he said.
Pulling back the curtain
In his book, Oscarmetrics — The Math Behind The Biggest Night In Hollywood, Zauzmer digs through every film category and asks lots of questions of his data.
What type of source material is best for an adapted screenplay win? Are big-budget films more likely to win best sound editing? How important are the awards given out by the various guilds in determining their category winners? At what age are you most likely to win in the four acting categories?
Some of the results are surprising, such as what he found when he started sifting through past Oscar nights to find which category was the best predictor of a best picture winner.
“One thing that is extremely important is looking at the full landscape of the other nominations, because these categories don’t come in a vacuum,” Zauzmer explained.
For example, a best picture winner ideally comes with nominations in the director and screenplay categories.
But Zauzmer’s research also found a listing in the best film editing category is a better indicator of a best picture win.
There are exceptions — Birdman, which was filmed to look like a single take, won best picture in 2014 without a nomination for editing, but that is the only film since 1981 to achieve that feat.
This very factor could hurt Sam Mendes’ real-time war film 1917, which doesn’t have a best film editing nomination for similar reasons to Birdman.
But the rest of the data is still pointing to a best picture Oscar for 1917.
“1917 is the likely frontrunner for best picture with a Producers Guild win, a Directors Guild win, a Golden Globe win, and by that same token, the frontrunner for best director,” Zauzmer said.
“There has been a bit of a separation between [the best film and best director] categories in the last few years but, historically, those two categories are still pretty tied, and the math likes Sam Mendes for [best] director.
“For the acting awards, there have been four actors that have really swept through awards season — Joaquin Phoenix for Joker, Renee Zellweger for Judy, Brad Pitt for Once Upon A Time In Hollywood, and Laura Dern for Marriage Story.”
Cue the music
Zauzmer said there were “factors that can’t be captured by a computer, that a smart Oscar predictor will use in their non-mathematical predictions”.
“A really good example is if there’s any sort of news story or social movement or sentimentality that comes really close to the end of Oscar voting,” he said.
“Everything I’m including in my model — things like BAFTAs or Golden Globes — all come fairly early in Oscar season in terms of when their deadlines are to vote.
“If there’s something out there, like a sentimental favourite that affects all of those awards in a row, then maths can pick that up just fine.
“But if there’s something that happens at the very end, some late-breaking news, that’s not something maths is going to know about.”
He said the changing film landscape meant factors relevant when the Oscars started in the 1920s, or trends even as recent as the 2000s might be unimportant now.
“There are all sorts of stats where someone could say such-and-such has never happened [such as] no foreign film has ever won best picture, no Netflix film has ever won best picture,” Zauzmer said.
“We could wake up [after this weekend’s ceremony] and that sentence is no longer true.”
In a neat ending, worthy of a movie, Zauzmer now works in the very field that helped inspire his Oscar-predicting sideline.
In 2015, after finishing at Harvard, he was hired by the LA Dodgers baseball team to work as a statistician — yep, he became one of the Moneyball guys.
But on Sunday (LA time) he’ll be putting baseball away to see if he can hit a mathematical home run with his other passion.