Can math predict the success of film?

In the 'Numbers' TV-series they can even predict the shape of your neighbours tird 2 months from now ;)

In the Netherlands we have a rockband that was always trying to get big in the rest of the world.
For their second or third album they worked with a producer/composer who knew 'everything' about music.
The singer said in interviews they used statistics to place the chorus on the right spot to make it a hit.

Guess what?

You never heard of them, unless you live in The Netherlands of Belgium :P
This 'study' doesn't seem to say anything new - it's merely proving that the more you spend on advertising dollars, the more people are going to see your film.

It's that simple. If you're going to spend $5 advertising your film, then there's no way you're going to have as many people seeing it (and therefore not as much takings) as say, Avatar.

The whole study is actually nothing new, it's essentially an ROI chart that just accounts for more films. Studios do this all the time - that's how they work out how much money to spend on advertising. It's not just a 'oh I think we'll spend x amount, because I plucked that number out of my head', they've done the charts on return on investment - ie advertising dollars vs box office takings.
Last edited:
BTW, interesting article.
It says what a lot of people in here say as well: marketing is the key to selling a movie.
Brain-position is the key to word of mouth.

I wonder how they put in the soft data (word-of-mouth in real life of private emails) next to the hard data (money spent on marketing, searchresults and analytics on blogs).


IndieTalk's Resident Guru
To many accountants and math people in the business already.

Fun article, but nothing new and nothing that can actually work.
If it did, and there was the "right"number then every film a studio
released would be a huge success. No studio would ever spend the
wrong number.

Interesting that the "John Carter" numbers aren't included.
Interesting that the "John Carter" numbers aren't included.

Probably because that case didn't support the theory.
In The Netherlands we've had 2 famous sociology scientist who got caught inventing statistics to proof their theory (Stapel) or excluding data that didn't support the theory.

Stapel was famous numerous publications, one of those was about 'discovering' that "meateaters are less social than vegetarians".
It turned out he spend a few evenings filling in 100s of questionaires by himself. :lol:
This 'study' doesn't seem to say anything new - it's merely proving that the more you spend on advertising dollars, the more people are going to see your film.

According to the math, it is correct.

$$$ in advertising = number of people seeing the ad = the number of people who will then spend the money to do what your ad suggests (see the movie, buy the dvd, buy the merchandise, etc).

Of course this math is developed by the ad agents who you will be paying to do that advertising.

There are several examples of people using social media (facebook/twitter/google +) to spread the word without the need for a traditional advertising outlet.

Their math cannot predict word of mouth which remains the best advertising and can trump any amount of money spent in advertising.
I have two concerns about mathematical modelling:

1) The first is that I don't see how it can tell a good story from a bad one that is identical. Conan the Barbarian did well under Arnold in 1982, and it in fact opened the door for his film career; but the other ones with other actors have since failed, even though they were telling the same story.

2) The second is the fact that numbers are not the whole story (pun intended). Numbers must always be read in context, and, if the researchers don't know the context, they cannot interpret the numbers properly.
The business is more based on what you have done before to sway an investor such as a bank and a distributor to put money up front. Anyone remember how fast The War Of The Worlds got funded when Steven Speilberg was attached as the director and Tom Cruise was attached as the main star?
I agree with what a DP I recently worked with said about getting a bank loan to make a movie. If you can show them documentation that a cable tv network will air a film you are looking to make,the bank will give you a loan without putting a lean on any property you own because the network agreeing to pay a set fee to license showing your film is how the bank will get the loan repaid.
While I believe you can use math for that, I don't see it as practical. The variables can change wildly and there isn't enough relevant data that is up to date to allow it to happen with that much accuracy.

Nice list of 3 movies. I didn't enjoy any of those.... and all of them are smack bang in the center of my preferences. Not my demographics, just my preferences.


IndieTalk's Resident Guru
The race is now on
You’re a little over 3 years behind the times on this. First prize
was awarded on Sept 2009 and the Business article you link
is also Sept. 2009.
"In the second contest, the company will award two $500,000 prizes, a preliminary award after six months, the other after a year and a half. "
So the second contest ended March 2010 and March 2011. Do
you know who won?

This system does not suggest how successful a movie will be. It
improves the accuracy of predicting what movie someone might
enjoy based on what they have enjoyed in the past.

Using your John Carter example - it’s quite clear that many people
enjoy sci-fi/fantasy/action movies. So it’s clear than many people
would enjoy “John Carter”. But just because many people like
sci-fi/fantasy/action movies doesn’t mean they will pay to see all
sci-fi/fantasy/action movies that are released. So they wait until
the movie is released on Netflix rather than pay to see it in a
theater or buy a copy.

What people may enjoy and what they actually buy are very different.

There may be some predictability, if one algorithm is better than the other, but we'll have to see. I wonder what those algorithms would have said about John Carter - or Battleship beating Hunger Games in DVD sales.
Hunger Games - 5,777,664 units, $100,953,240
Battleship - 689,650 units, $12,649,359
John Carter - 1,059,463, units $18,470,437

John Carter sold 526,928 ($8,952,507) units in its first week
Hunger Games sold 2,176,055 ($36,971,174) units in its first week

John Carter dropped out of the top 30 9 weeks later with 1,059,463 ($18,470,437) units sold.
Hunger Games is ranked #6 this week and selling strong. 5,777,664 ($100,953,240) units sold to date.
I started this thread showing that, initially, Battleship beat Hunger Games.

That said, Rik, you drew a distinction between what people enjoy and what they would buy. But, if an algorithm can predict what people would enjoy, it would be a big step towards finding a more advanced version that would predict what people would buy.

I am not saying it's possible; I'm saying that your distinction, while important, MAY (emphasis may) not be an argument towards looking for an algorithm that can predict the success of movies. As of now, I'm also a skeptic, but I'll keep an open mind.
Last edited: