About
Ben Zauzmer, a Harvard sophomore with a long-time passion for math and movies, combined his twin interests to produce Oscar Forecast. By devising formulas designed to calculate the chances each movie wins Academy Awards, he offers a mathematical approach to the exciting and difficult art of Oscar predictions. Last year, his model correctly predicted all eight of the major categories. Please feel free to e-mail bzauzmer@college.harvard.edu with any feedback.
For each category, Ben compiled all of the significant award shows (e.g. BAFTAs), the guild awards (e.g. Writers’ Guild Awards), any corresponding Oscar nominations (e.g. Best Sound Editing and Best Sound Mixing). and Metacritic and Rotten Tomatoes scores. With all of these numbers in a chart for each nominee in the category over the past 15 years, using a formula from statistics, Ben derived the best approximation of the relative factors of each award and critic score. The weights created by this model are the ones that would have most accurately predicted each of the past 15 years. Ben only used math; no personal hunches were involved. These factors were applied to this year’s nominees – one formula for each category – and the percentage was calculated as a movie’s score out of the total scores.
Of course, the Oscars are a human endeavor, so they cannot be modeled perfectly by mathematics. The Academy of Motion Picture Arts and Sciences is just that – a combination of art and science. These predictions may prove to be highly accurate, but indicators alone cannot guarantee any results. For example, in 2009 this method would have predicted 19/20 awards, but only 16/20 in 2010.
A few notes about the percentages listed on the predictions page:
- The minimum score for a movie was set to 1%. All values were rounded to the nearest percent. Finally, the values were rounded to force the total to 100%.
- For most of these categories, the data used goes back to the 69th Academy Awards in 1996. The exceptions are Best Score (data since 1999), Best Animated Feature (2005), Best Documentary – Feature (2002), and Best Visual Effects (2002).
- All Oscars are covered except the three awards for short films. There is not enough data online to make accurate predictions for these three categories.
- All references to the Academy Awards and the Oscars are copyrighted property of the Academy of Motion Picture Arts and Sciences. For more information, please see http://www.oscars.org/legal/regulations.html#copyrights. This site is for editorial use only and is not intended to earn money. The photograph on this website is protected under ©A.M.P.A.S.®.
Ben's Oscar Forecast 
[…] though, Harvard student Ben Zauzmer has developed a formula for calculating who will walk away with a statuette this Sunday and posted those math-based Oscar […]
Ben,
Thank you for helping me win my Oscar pool. I have been going to the same party for the last 20 years and have won the pool more than any other guests over the years. Your stats this year helped me garner another victory. As the wife of an electrical engineer/computer guy (Princeton ’90), I am all for the numbers. (Love Nate Silvers and “Moneyball.”) Thank you again and good luck to you in your studies. I am a geriatrician and should you need one in Los Angeles, please feel free to contact me.
Be well, Sue
When calculating the rottentomatoes percentage, are you using the overall percentage or average rating? And how does that factor in exactly? For example:
The Artist – 191 reviews – 97% fresh – 8.8 average rating
The Muppets – 190 reviews – 96% fresh – 7.9 average rating
It would appear that The Artist would have the advantage since all things being equal it had the higher average rating. And what if it had fewer reviews?
I enjoy your theory and will check back on Sunday. Nice work.
Hi Ted,
Thanks for your comment.
I used the overall percentage for Rotten Tomatoes, and the average critic score for Metacritic. I did not take into account the number of reviews, since both of the figures are already averages. For some awards, there was so little correlation between critic scores and the results that it did not matter much, but for other awards the higher-rated movies actually did tend to fare better over the past decade.
Happy Oscar weekend!
Ben
The brief about section alluded to results from 2009 and 2010. Have you taken this formula back further? If so, did you also come up with an average +/- for these catagories to a ratio that could coincide with your prediction? For Example: Best Picture – The Artist (28% at 98% accuracy)
The formula was based on 2001-2010 for most categories. So, there would be similarly high results for any of those 10 years. The real test will be 2011, which is not part of the data but part of the predictions. Should be interesting to see how accurate (or not) math can be in this field.
Coming up with a margin error would require having previous years experience to know how far these results deviate from the actual awards. So, if this continues for a few more years, I will be able to calculate that information.
2/23/12
Ben:
Please promise that you will use your amazing intellect only for good!!!
–Steve Stone
PS–I know you will!
Sounds good to me. It’s a promise. Thanks for checking out the site.
Hey I checked out this site last year and I am back for more this year! Please post your predictions. Thanks!
Thanks for checking out the site. Many of the variables used in these predictions involve results from other award shows leading up to the Oscars. So, the predictions will be released immediately following the last set of awards – the Costume Designers Guild awards on February 19, the Tuesday before the Oscars.
This is exactly the type of website I’ve been looking for on the internet for years, so thank you for putting this all together! Question for you. Cinematography seems to be the category that has the least amount of certainty as to who the winner would be statistically. I’m assuming Deakins is at the top of the list because of his win at the ASCs. Were there any other factors that went into that category? And do you think, non-statistically speaking, that his lack of a win at the Oscars for so long can only help his chances?
Thanks!
Thanks for checking out the site! I’m glad you enjoy it.
Cinematogrpahy is definitely the tightest race of the night. The factors that helped Lincoln and Life of Pi were the Production Design nominations and the total nomination numbers. There were other factors as well, but those are the significant ones.
I did a blog post (https://oscarforecast.wordpress.com/2013/02/02/oscar-veteran-actors-and-directors/) about Oscar veterans being helped out of sympathy. The answer for actors is “no,” and for everybody else there just isn’t a large enough sample size to say for sure. But on an anecdotal level, yes, Deakins could get a small boost from his past record. But I’m not in the business of factoring that into my predictions.
This will definitely be my toughest choice. Thanks for the input!
what statistical method did you use to come up with the weighting? thx – like the site!
Thanks for checking out the site! Glad you like it.
For stastics people, it’s simply a series of linear multiple regressions. For non-statistics people, I created a formula for each category that best reflects how important different indicators are over the past 15 years.
If you right about your director choice, you are a genius. (I thought Hollywood loved Spielberg.)
Have you ever listed your personal choices vs. predictions?
So far, I’ve avoided that to make sure I don’t confuse anyone as to which picks are mathematical and which involve my own opinion. But as a movie buff and not just a numbers guy, I can assure you I have plenty of my own opinions, even if I don’t factor them into the predictions on this site.