Nate Silver predicts an Oscar upset!

This article was originally on a blog post platform and may be missing photos, graphics or links. See About archive blog posts.

If all you do is read the trades, you probably have no idea who Nate Silver is. But if you’re a baseball fan or a political junkie, you know that Silver is the most brilliant statistical guru of our time. The baby-faced statistician came to fame as the guy who developed the PECOTA system for forecasting baseball players’ career development and performance, earning the attention of general managers when it became clear he knew more about their players’ potential than they did. Last year, Silver switched to politics, where he was the first soothsayer to take Barack Obama seriously when his peers were all lining up behind Hillary Clinton. After starting the ultra-cool FiveThirtyEight site, Silver accurately called every U.S. Senate race and 49 of 50 states in the presidential race (he missed Indiana).

So if Silver was right about Obama, could he be right about Mickey Rourke? New York magazine asked the Chicago-based numbers geek to work his statistical magic with the Oscars. I have to say--as someone who thinks the hapless Gurus of Gold pundits are about as persuasive as a Kennedy assassination enthusiast--the results are fascinating. Silver says he used a process called ‘logistic regression’ to analyze a 30-year database of Oscar history, studying everything from the films’ genre, MPAA classification and opening weekend box office to whether someone benefited from being nominated in another category. Some things mattered not (MPAA classification), some mattered a lot (the academy roundly ignores the comedy genre).

So here are a few of Silver’s picks--with some of the justification for his choices. As with his political calls, he gives each pick a percentage of accuracy:

Lead actor: Mickey Rourke: 71.1%, Sean Penn: 19%. The Call: Since Rourke and Penn split the two awards that traditionally predict success in the category (SAG and Golden Globes), he gives the advantage to Rourke, since once an actor wins--as Penn did five years ago--his odds go way down, while someone who’s been nominated without winning sees their odds increase.

Lead actress: Kate Winslet: 67.6%, Meryl Streep: 32.4%. The Call: Streep has won some big awards (including SAG) but she’s already taken home Oscars, while Winslet is ‘sitting on her sixth nod without a win.’

Best picture: ‘Slumdog Millionaire’: 99%. The Call: It’s a rout.

Director: Danny Boyle: 99.7%. The Call: This is where the academy rewards ‘edgy’ films like ‘Slumdog’ (Ang Lee for ‘Brokeback,’ Steven Soderbergh for ‘Traffic’). So Boyle is a shoo-in.

Supporting actor: Heath Ledger: 88.5%, Josh Brolin: 5%. The Call: It’s a lock.

Supporting actress: Taraji P. Henson: 51%, Penelope Cruz: 24.6%, Viola Davis: 11.6%. The Call: A huge upset. It’s the hardest call of all, since most of the major supporting actress awards were won by Kate Winslet, which, as Silver puts it, ‘is not so nice for our computer.’ He says Cruz would be the logical default, but his computer says that ‘Benjamin Button,’ which looks like a shutout everywhere else, ‘ is the only best picture nominee with a supporting actress nod, and best picture nominees tend to have an edge in the other categories.’

All I can say is that if Silver gets that call right, next year he’s going to have to predict all 108 Grammy categories. Politics may be easier than showbiz, at least when it comes to reading the minds of voters.