Description image

Could an Equation Determine Whether a Movie Will Be Successful?

01.22.13 @ 1:30AM Tags : ,

The movie industry is no stranger to equations and analytics, and many of the decisions they make are based on such numbers. It’s very often a gamble, however, because no system is perfect in its predictions, and there is no guaranteed way to anticipate many of the other factors that go into these releases. But what if you could develop a mathematical model that could accurately predict how successful a movie would be? That’s what the Tottori University Department of Engineering in Japan is attempting to do with their own equation. Check out the video below:

Here is another look at the equation (click for a larger view):

Check out their simulation versus the actual release:

While this could be great for Hollywood executives to figure out whether they should be taking a loss on a film that hasn’t even been released yet (or popping champagne corks), it doesn’t necessarily help independent filmmakers all that much unless they are doing a ton of advertising and promotion — and actually releasing in a traditional manner. Analytics are exceptionally fascinating to me, and it will be interesting to actually see this implemented and what effect it might have on the industry.

The downside to equations like this, of course, is that it may put more emphasis on projects that are guaranteed to make money, rather than those that may be pushing the boundaries of mainstream cinema.

What do you think? Have you ever seen equations used before to determine how successful a film might be? How could you see independent filmmakers utilizing the power of the internet and social media to determine how successful an online distribution plan may be?

[via DigInfo & FilmmakerIQ]


We’re all here for the same reason: to better ourselves as writers, directors, cinematographers, producers, photographers... whatever our creative pursuit. Criticism is valuable as long as it is constructive, but personal attacks are grounds for deletion; you don't have to agree with us to learn something. We’re all here to help each other, so thank you for adding to the conversation!

Description image 22 COMMENTS

  • He’s Asian, he must be right.

  • joe marine, it´s not totally off-topic, since it´s about measuring how movies affect viewers, this paper may be of your interest, if you not already read it.

    it´s about the presence of pink noise (1/f) in the editing, montage, of main stream north-american movies and it evolution in time, with more and more movies using it, mostly unconsciously, if not all.

    There is other papers that suggest that 1/f noise is the pleasant pattern for temporal punctuation, be it in music or in movies (since the space between cuts is like the musical intervals in film sequences).
    Attention and the Evolution
    of Hollywood Film
    James E. Cutting
    , Jordan E. DeLong
    , and Christine E. Nothelfer
    Cornell University and
    University of California, Berkeley
    Reaction times exhibit a spectral patterning known as 1/f, and these patterns can be thought of as reflecting time-varying
    changes in attention. We investigated the shot structure of Hollywood films to determine if these same patterns are found. We
    parsed 150 films with release dates from 1935 to 2005 into their sequences of shots and then analyzed the pattern of shot
    lengths in each film. Autoregressive and power analyses showed that, across that span of 70 years, shots became increasingly
    more correlated in length with their neighbors and created power spectra approaching 1/f. We suggest, as have others, that
    1/f patterns reflect world structure and mental process. Moreover, a 1/f temporal shot structure may help harness observers’
    attention to the narrative of a film.
    attention, cinema, film, visual momentum, 1/f

  • on 01.22.13 @ 9:53AM

    It could be fascinating to see where this research leads, since I have a bunch of people coming to me and asking how a film will be profitable.

  • What do those graphs mean? horizontal is the number of days. vertical is the number of tickets? or money spent? How is this predictive in anyway. They’ve come up with an equation that says, if you advertise your movie a lot, people will go see it, or if people are talking about your movie a lot, people will go see it. You need math to tell you that? These graphs don’t tell us anything. And how does it account for the actual quality of the work. It’s incredibly unclear what information they are giving us here. From what they’ve given us I could draw the conclusion that they are the smartest people in the world, or that they drew some colored lines on a paper and said, “see. They match.”

    There’s nothing in the video that says the equation tells you how to get people talking about your movie, or that a certain movie will get people talking, but simply that people talking and advertising lead to more sales. Again, we all knew that.

    Am I missing something?

  • Did I miss something? Of the three factors, only external influence (marketing) can be predicted. A studio knows what it will spend marketing and how many stars it will get onto talk tv & radio. This hopefully creates a buzz which influences indirect communication.

    Direct communication remains the holy grail. If I can get every one of you to tell a friend to see my movie, I’m set. But that’s. Not going to come into play before release. You have to see it before you can recommend it.

    I fail to see the usefulness of a tool designed to predict which movies in theatres are going to be hits. As a producer, I want a tool to tell me months earlier, before I’ve sunk all my money in a flop.

    Over time, Hollywood seems to have given up on direct communication. Movies have much shorter theatrical runs, and Hollywood banks on big opening weekends.

  • Swami Digital on 01.22.13 @ 7:56PM

    The title is a bit misleading, “Could an Equation Determine Whether a Movie Will Be Successful?”

    The proper description would be that the equation seeks to describe what makes a movie successful. Mathematical models attempt to describe reality, not determine it. This is a not a grammar issue, but a key conceptual issue. Once they accurately capture, or model the behavior in the system that they seek to describe, they can be used for prediction, provided accurate initial conditions.

    • Right, that’s the eventual goal, to predict whether a film will be a hit before it is released.

    • “The proper description would be that the equation seeks to describe what makes a movie successful.”

      No, that’s actually a lot worse, because it confuses correlation with causation. Models, well, MODEL some phenomenon, and they are assessed scientifically by their predictive value. A model that proposes a mathematical relationship between two things does NOT need to describe a mechanism of action to be complete; the mathematical relationship is enough for the model to be tested scientifically.

      • Swami Digital on 01.25.13 @ 8:51AM

        You are actually incorrect, because simple correlation will mean that the model does not accurately describe the relationship, and thus will be unfit for prediction purposes. Perhaps this is a simple difference, but in the discipline of machine learning and predictive modeling it is important to discard variables that have a simple correlational correspondance because they destroy the model’s capability for generalization.

        • No, I’m not incorrect. Having extra variables that are correlated can, of course, violate independence assumptions of certain learning algorithms (Naive Bayes, for instance) but you’re trying to assert that they need to be discarded in a general sense, and this is simply not correct.

          In fact, one very famous example of machine learning is predicting a person’s age based on a sample of his or her writing. This is because age and writing style are correlated, but it is NOT the case that the model seeks to describe how the person’s writing MAKES, or causes, the person to have a particular age.

          If this doesn’t seem applicable, let’s get back to what you originally said: that “Could an Equation Determine Whether a Movie Will Be Successful?” is misleading and that “the equation seeks to describe what makes a movie successful” is better. This is not the case, because when it comes to predicting whether a movie will be successful, we don’t need to know what MAKES it successful; we just need to know which things correlate with its success. For instance, imagine that some kind of discussion on twitter is a really useful predictive feature: this does NOT mean that that discussion actually causes the movie to be successful. It MIGHT be the case (e.g. people see the movie because of the discussion), but it could also just be the case that people who are excited about the movie talk about it amongst themselves, and that if twitter were to crash for some reason, the same number of people would ultimately see the movie. This would be a perfect example of where a correlated, but not necessarily causative feature is useful for prediction.

          I normally wouldn’t go into such a lengthy diatribe, especially since I think your problem is mostly that you’re not expressing yourself properly, but the fact that you’re telling others that they have a fundamental misunderstanding when this is NOT THE CASE could confuse a lot of people. Predictive models are predictive models, period. There is absolutely no requirement that they have anything to say about causality or indeed that they have explanatory power.

  • There is a guy in Australia doing his PhD on this very topic. He was a script reader for a BIG producer and they used math and paraconsistent logic to “predetermine” success.

    Call ⊨ explosive if it validates {A , ¬A} ⊨ B for every A and B (ex contradictione quodlibet (ECQ))

    It’s real and it is already proving useful.

    • That fancy-looking logic in the second sentence just a somewhat awkward statement of the principal of explosion. Care to elaborate on how paraconsistent logic actually relates to making such a predictive model?

      • its not my PhD. So not I cannot. But isn’t it interesting that it is related to special effects!

  • The “Buzz” aspect of this seems culturally subjective. I’m sure that people overhear conversations that others are having on the street much more in Tokyo, where sidewalks are packed with people and public transportation is popular, than in Los Angeles, where most people drive.

  • The Marvel Comics and Transformer franchise successes seem to show that the crappier a film is, the more money it will make worldwide. Audiences have succumbed to the bigger, louder, dumber, crasser = “good” model of film making.

    Let the Lowest Common Denominator rein supreme!


    Whether this equation works or not isn’t really the issue, it’s whether or not people in power believe it works. This is extremely worrying to me and should be to you. Once powerful people (movie execs, music execs etc) think there is a template to success all films will be required to fit this template and therefore “Creativity” and “diversity” will be dead. We have already seen this in modern cinema but have been spared the total and utter domination of this cancer due to the uncertainty and lack of scientific evidence for a working formula, but should a mathematical equation ever be proven to have validity, and be accepted by the leading lights in the industry, we’re all screwed. You’ve only got to watch the Disney channel to see the kind of world we’ll be living in should this be allowed to happen.

    Urk,./,/. I feel sick…

  • This seems to be a formula to figure out successful marketing (advertising + direct word of mouth + indirect word of mouth) more so than the film itself. What this could mean is that marketers can better gauge success and figure out which aspect of the campaign needs additional resources. This could be a very good thing for indie films that have marketing budgeted in from the start.