Statsgasm Returns: 2015 Best Picture Voting Simulation

I am excited and honored to announce the official return of Awards Daily’s Statsgasm for the 2015–2016 Oscar season. As you may have noticed in some of Sasha’s critics awards previews recently, I have picked a few new magic tricks since you all last saw Statsgasm in full back in 2014 – specifically, my learning how to build graphs and dashboards in the wonderful visualization software package known as Tableau. I’m far from mastering Tableau’s many powers, but the skills I have picked up should prove sufficient in creating a more dynamic, interactive iteration of Statsgasm for you all.

With Oscar nominations less than 24 hours away, this first episode of Statsgasm 2016 will revisit and examine every Oscar prognosticator’s nightmare topic: the preferential voting system used by the Academy to determine their slate of nominees in many categories, including Best Picture. Like many other complicated subjects with esoteric terminology, I find that the best way to build understanding is often through learning by doing – with repeated experimentation and simulation whenever possible.

Let’s jump right into it, shall we? “Ready when you are, Sgt. Pembry.”

Since we do not have access to any AMPAS voting data (and we will likely never see that sort of information), we will need to rely on proxy ballots to perform this Best Picture voting simulation. (Dr.) Rob Y for many years has performed his Simulated Ballot for Awards Daily using direct input from you, the reader. He has just completed the simulation for this year, which may be accessed here.

For Statsgasm, I have used end-of-the-year Top 10 lists submitted by film critics, bloggers, journalists, filmmakers, etc. There are obvious limitations to this methodology, particularly with regards to sample size (500 voters vs. 6000 voters) as well as the simple fact that critics/bloggers are a completely different species from AMPAS and other industry voters. Hence, we must keep in mind that this voting simulation isn’t really meant to predict how AMPAS voters will behave; rather, it’s merely meant to demonstrate the nuts and bolts of how preferential voting is performed, and possibly illuminate why certain outcomes occur.

Like Dr. Rob, I simulate the Best Picture voting process in three stages. I won’t go into detail here as these rounds are explained in the Tableau dashboard I have created, but if you would like a more thorough explanation, please use the 2014 simulation as a reference. Navigating the dashboard should be straightforward – just scroll through the tabs on the very top to move between pages.

So this year, our simulation resulted in seven BP nominees:

The Assassin
Brooklyn
Carol
Inside Out
Mad Max: Fury Road
Son of Saul
Spotlight

A few observations from the results:

• Once again, the simulation demonstrates that #1 votes are critical in landing a Best Picture nomination. For example, The Assassin showed up on 71 of 500 ballots, but since it managed to be ranked first on 20 of these ballots, it was nominated for Best Picture while Ex Machina (which had twice as many overall mentions as The Assassin but only 8 first place votes) was not. Consensus ultimately rules the day when determining winners, but strong passion can easily carry a film to a nomination.

• Based on the outcomes from the previous voting exercises conducted by Dr. Rob and myself, I theorized that it appears a film needs to target at least 3% of the first round votes in order to have a shot at the nomination. That 3% theory seems to hold true again in this simulation, as all seven BP nominees did obtain at least 15 first place votes. It appears that obtaining more than 2% of votes through the transfers that take place in Rounds 2 and 3 to reach the 5% nomination level is too much of an obstacle to overcome for most films.

• Mad Max: Fury Road was the runaway winner in terms of overall mentions and #1 placements. Unsure if AMPAS will be as passionate for it, but “Witness!”

For a more wonkier discussion that includes reference to the “Goldilocks Distribution” that is needed to obtain the ever-elusive 10 Best Picture nominee lineup under the current system, I again invite you to take a look at Rob Y’s work.

This concludes Statsgasm’s 2015 Best Picture voting simulation. Please feel free to ask any questions in the comments or email me directly! And stay tuned for more Statsgasm posts (including the inevitable return of the Oscar prediction models) in the coming weeks.