Fact Check March #1

@fourthwall · Fourth Wall
Apr 17, 2019, 8:04am PDTEdited

For this first factcheck, I decided to go with a classic from Furiosa and an often repeated statement on FML Discord.

Theater counts determine the (BO) ceiling, not the (BO) floor.

This fact is used typically when a(n awful looking) movie opens in a significantly high number of theaters. But is it true?


Theoretical aspect

From a theoretical aspect this statement is 100% true. A movie can be in 5,000 theaters, that is no guarantee that people will actually show up. So regardless, of the theater count, a movie can still make $0 over the weekend. Theoretically. In practice, even the worst wide opener still managed to get $128 per theater (https://www.boxofficemojo.com/movies/?id=proudamerican.htm)

The other part of the statement is also theoretically true. Theater counts do determine the ceiling. Theoretically, again, you could screen one movie on all your screens, aim for the maximum amount of showtimes and be at 100% capacity for each showing. Assuming, maximum shows/screens/capacity, the only remaining factor for the total box office is the number of theaters doing this. I believe Phil once said he considers $1,000 per showtime the maximum. The highest average amount of showtimes per theater was probably either Star Wars: The Force Awakens or Avengers: Infinity War. Unfortunately, the showtimes report wasn't available yet for TFA and broke down for AIW. Digging through my archives I did find back the numbers for AIW, which apparently had between 15 and 20 showtimes per theater (depending on the math for PLF). Considering this, it is no surprise that TFA and AIW top the Wide Release PTA ranking with $59,982 and $57,599 respectively.

So in theory, the statement is TRUE.


Data Driven Approach

But what about the numbers? Because theory is one thing, it is reality we are interested in. We can all imagine that the more theaters a movie is in, the bigger the odds of people actually - for whatever reason - being in the theater for that movie. Perhaps, if Proud American was in twice as many theaters, twice as many people would have accidentally made their way into the movie. While its PTA would still be as abyssal as currently, its BO number would have doubled.
Let's have a look at a simple scatterplot using theater count and box office for the weekend starting with wide releases (theater count over 600). [https://public.tableau.com/profile/fourthwallfml#!/vizhome/FWFactCheckMarch/Dashboard1 - filter on WR in the Quick Filters].

Two things become clear:

  1. There is definitely an effect (unsurprisingly) between box office and theater count
  2. Both the ceiling and the floor stay relatively flat for a very long portion of the theater ranges. It is only at 3,400 theaters that we can discern a little lift a the bottom and at 2,800 a climb at the ceiling.

The slowly climbing floor does seem to give weight to our statement, i.e. no matter how high the theater count, the floor is pretty stable. But, graphs are easy to manipulate. When focusing on the floor by filtering the box office between 11M and 40M, we see the floor rising at around 2,400 theaters. [https://public.tableau.com/profile/fourthwallfml#!/vizhome/FWFactCheckMarch/Dashboard1 - filter on WR in the Quick Filters + BO between 11M and 40M]


8 Replies

  • Reply #1

    Interestingly, following the bottom line of movies going down from 4,300 theaters to 2,800, we find these illustrious examples: Lego Movie 2, Smallfoot, Lego Ninjago, Nut Job 2, Paddington 2, Sherlock Gnomes, The Kid Who Would Be King, Transporter Refueled, Show Dogs, Darkest Minds, Fun Size, Father Figures and Victor Frankenstein.
    Limiting the graph to wide releases between 1 and 2,300 theaters shows this flat bottom line. Moving away from the visual analysis, and figuring out a trend line gives a number of significant possibilities. The most obvious is the exponential one but polynomial and linear are also statistically significant. The R² was actually pretty good but R² values can often be deceiving. In order to make sure our "model" fits the dataset, we also need to look at the residuals or the errors between the predicted value and the actual outcome. A good model should result in random residuals. Alas, this is hardly the case here - although you will have to trust my word on it - with residuals clearly showing a pattern.


    So, scratch the trend line models. As a final analysis, I binned all observations based on their theater count. I divided the theater count by 100 and rounded to the nearest integer. In other words, movies in 860 theaters and 910 theaters both wound up in the same bin. For each bin, I calculated the minimum, 5% percentile (floor), maximum, 95% percentile (ceiling), the average and the standard deviation. As we found in the visual analysis, the floor and 5% stay pretty stable, but eventually the floor start to climb, albeit at a much slower pace than the ceiling. Perhaps one of the things were are most interested in is not the minimum but the 5% and 95% percentile. Basically what we are saying here that for any theater bin, we have a 95% confidence that our movie will end up between the red and green line. Deadlinesque ranges, without a doubt. But you can still see the effect.

    Gonna do a little Anthony here: Graph Coming.

    Other filters

    So far I've only touched on Wide Releases (opening weekends).
    But what about limited releases? [Same link - filter on LR in the Quick Filters]. Not much to say about it, except that it is pretty much a crapshoot or at least not driven through the TC.
    What about holdovers? [Same link - filter on Holdovers in the Quick Filters]. Compared to the wide releases graph, the "cone" is getting smashed together a little bit. Which makes sense because movies typically don't lose theaters in week 2, but they do lose quite a bit of BO.
    What about by genre? Well, this is actually quite interesting…Action / Adventure seems to do a lot better than the average per bin across all genres but that relates to the fact that the whole of MCU is Action / Adventure of course. What is interesting is that "Action" tends to overperform in bins 35-37 (think Geostorm, Deadpool, F&F franchise and Bond) but then underperform in larger bins. (Slight nuance is in order, these are typically older movies when that TC range was still like blockbuster territory. Nowadays we seem to have pushed that TC bar up a notch).
    Another fact that can be learned is to be wary of "Animation" in a lot of theaters as proven time and time again and already evidenced by the list of movies making the floor. Anything over 3,800 TC seems to go under the average performance in those theater bins. Which begs the question. Why do studios release animated fare in that many theaters? Is it really worth it?

    @fourthwall · Fourth Wall
    Apr 17, 2019, 8:04am PDTEdited
  • Reply #2

    Another little snippet can be seen on that "Comedy" line around bin 23, while we are dealing with a small sample here (there is only one Comedy in that bin 23), a filtering on this [Same Link - filter on WR in the Quick Filters + TC between 2,000 and 2,500] and a little walk through the top dots immediately tells you something. Quite a lot of these top dogs are African American targeted movies having a targeted release in a relatively small number of theaters. There was some discussion on this on Discord why movies like this aren't put into more theaters. And if this would actually improve their box office or not.

    I've come to the conclusion of my first fact check. Almost 3 weeks over due but I don't care. I set my own deadlines.

    Let me summarize:

    Theater counts determine the (BO) ceiling, not the (BO) floor.

    • While theoretically true, in practice theater counts do push up the floor.
    • Especially going over 2,400 theaters.
    • With regard to the ceiling, the key element isn't theaters as much as showtimes.
    • Animation has a scary low floor
    • AA targeted releases can do really well despite the lower TC range.

    In other words, the statement is only partially true.

    That concludes my first Fact Check.


    How about...

    Let me know what you think and suggest new things to verify!

    @fourthwall · Fourth Wall
    Apr 17, 2019, 8:17am PDTEdited
  • Reply #3


    Excellent work!

    For your next task - I challenge you to verify something near and dear to my heart, and something which will be appropriate for next week:

    THE D.D.I.E. - The Disney Drive In Effect.

    There are Drive In effects, and then there is an entirely different animal, the Disney Drive In Effect, which I am convinced operates above the level of any other Drive In Effect.

    Disney Accounting at its finest!


    @bswhite · BSwhite 1970's Discoplex
    Apr 17, 2019, 8:30am PDT
  • Reply #4

    I have a basic understanding of thearter counts so far and I've always seen this as the ceiling and not the floor. A movie's floor has to be determined by different method or formula. This could be done by survey. In which you ran a website that shows percentages of which movie people were likely to see over such and such a weekend. The floor of a movie is definately hard to predict. There's weather factors, time factors, popularity factors, and different factors that can ultimately cause drastic deviations from the mean. If you have a 3,500 tc for a movie you would see that max value rise. Whereas the floor of a movie is going to be entirely different bc there could be many empty seats. The only thing you can really hope for in predicting the floor is to get ahold of the marketing plan for a movie and see how much exposure were fans given to a preview.

    @zman2018 · Terminus
    Apr 17, 2019, 8:59am PDT
  • Reply #5

    Fourth: This is a really excellent analysis more thoughts later.

    Mr. White, It's just called a drive in effect. All studios do it. Disney is also not the only studio with some interesting accounting. So you're right but for the wrong reasons; your singular talent.

    @realfakefilms · Real Fake Films: Chunnel △
    Apr 17, 2019, 9:01am PDT
  • Reply #6

    Great analysis 4th! I'd love to see you take down the Rule of 3 - that is a bunch of alternative facts lol

    @banejo · Naturalized Bane-Jo Sudswater
    Apr 17, 2019, 9:04am PDT
  • Reply #7

    Great analysis 4th! I'd love to see you take down the Rule of 3 - that is a bunch of alternative facts lol

    He did that last year:


    @seesomething · IYSSSS (IOU)
    Apr 17, 2019, 9:07am PDT
  • Reply #8

    Ummm..can I just have more Robert Downey junior pics please?

    @user4dx6kp · andra79's Cineplex
    Apr 18, 2019, 11:32pm PDT
Add Reply