Oscar By The Numbers: Time To Tweak the Model?

On Oscar nomination day this year, I confidently predicted in this post that The Imitation Game would capture Best Picture honors this year, based on a mathematical model I developed some years ago, and which is described more fully in the link.

I was obviously wrong: Birdman took the title instead, as many pundits predicted it would. So is it time for me to scrap my own model?

I don’t think so. Let’s review what the model predicted in the first hours after the nominees were announced, before all the ensuing media kerfuffle. Here’s the key quoted segment from the original article, with the numbers:

So what happens when you load this year’s Best Picture nominees into the database and crunch the numbers? You get these results:

  • The Imitation Game: 59.7%
  • Birdman: 57.2%
  • The Grand Budapest Hotel: 52.3%
  • Boyhood: 41.3%
  • American Sniper: 34.9%
  • Whiplash: 33.5%
  • The Theory of Everything: 29.9%
  • Selma: 0.9%

That’s a pretty tight contest between the top three films, and I’m somewhat pleased and gratified to see that Boyhood is not among the leaders, even though as I type this people are gushing about it as the shoo-in favorite after its Golden Globes performance.

The model did identify Birdman as a tight second place finisher, so that was not too far off the mark. The model also predicted that Boyhood did not really have a chance at the big prize — at a point in time when “everyone” had already earmarked it as the year’s favorite. I’m pleased with that result, too.

From a bettor’s standpoint, “Best Picture” is obviously a win or lose, all or nothing prospect, and so in that way, the model failed. But it didn’t fail by much, with only a 1.5% prescriptive difference between my pick and the actually winner, and the presumed favorite (at the time) accurately relegated to also-ran status. So I don’t think that it’s ready for the scrap heap quite yet.

I do, however, think I need to figure out a way to factor in one significant change in the nominating and award process that occurred after I developed the model: the new opportunity for the number of Best Picture nominations to be greater than the number of all of the other categories. This must have an impact on the way that the historical five-to-five correlations map onto recent and current years, though I have to figure out exactly what that impact might be, and how to capture it.

But I think that’s a minor tweak, at most, and not a full revamp. So watch this space on Oscar Nominating Day 2016, when I’ll be back to boldly call it again — and hopefully won’t lose another squeaker!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s