Skip to content


In “High Fidelity”, Rob Gordon asks the profound question: “Did I listen to pop music because I was miserable, or was I miserable because I listened to pop music?” The same chicken-and-egg logic applies to film box office performance: are films successful because we buy tickets, or do we buy tickets because films are successful?

Here’s a graph that supports the second claim – we buy more tickets when a film is successful, or in other words: success drives even more success:

Exponential Box Office Vs Rank 2002-2013

If tickets were bought based solely on each individual’s personal and isolated taste preferences – we would have expected this plot to be linear – meaning box office would increase proportionately with rank. But this plot is clearly non-linear. It looks logarithmic.

So it seems that film box office has a recursive nature, where success drives even more success in a positive feedback mechanism: the difference between rank 50 and 60 is much bigger than the difference between 60 and 70. I call it “The Rob Gordon Effect” (ironic, because this character was kind of a loser).

There are many possible explanations, but I think if we’re honest with ourselves, we do make decisions based on news of success or failure: a breakout box office phenomenon is an “event” we all want to be part of. Nobody wants to be one of those people who haven’t seen Avatar, and no one wants to be mocked at the office for being the only person who was so out of touch to actually pay to see Johnny Depp’s “Lone Ranger”. Everyone wants to be associated with the winners, making it very hard for perceived underdogs to triumph. It’s not very romantic, but it’s true.


Ok, that was part 1. From here on, it’s all unnecessary brain drilling.

So far, we’ve established what could have been a perfectly adequate blog post all on its own. But if you are a truly brave nerd, here is a slightly more complex elaboration:

The above plot is actually not perfectly logarithmic. Upon closer inspection, we see that the log curve really starts only around rank 300. I know what you’re thinking – this can be a tricky observation, as the damping of the curve may be mistaken for linearity. Therefore, here are the four normalized segments of the curves: ranks 1-300; ranks 301-600; ranks 601-900; and ranks 901-1,131, with the box office grosses normalized on a range of 0-100, so that we can objectively compare the curvature:

1. For real blockbusters (over $120MM domestic gross), the recursive effect is strong, meaning blockbusters get a disproportionate extra boost in the box office by climbing up the ranks:

Box Office Vs Rank 2002-2013: ranks 1-300

2. “Almost-blockbusters” ($70MM-$120MM domestic grosses) are less affected by this recursive effect: they are much less logarithmically organized, but not perfectly linear either:

Box Office Vs Rank 2002-2013: ranks 301-600

3. And finally, smaller movies (ranks 601-900 and 901-1,131; less than $70MM domestic grosses) are almost perfectly linear:

Box Office Vs Rank 2002-2013: ranks 601-900

Box Office Vs Rank 2002-2013: ranks 901-1131

The conclusion:

The conclusion of all this is that the “Rob Gordon Effect” really kicks in only for mega blockbusters (over $120MM domestic gross): they are disproportionately boosted by news of their relative success. But smaller films are in fact NOT affected by such news (especially under $70MM domestic gross). Therefore – there is a threshold above which a movie gets turbo-charged by its own success. The practical takeaway for studio execs is: if your movie has a chance of breaking that threshold ($120MM as of March 2014) – you should give it that extra marketing push. It will pay off.

* Raw B.O. data from (1,131 films total).

– Like this post? Follow me on twitter: @stavjdavis and/or like the Facebook page


Recently, it feels like movies are getting longer and longer. Wolf, Smaug, Ranger, Fire, Prisoners, Superman and Gatsby – all clocked in at over 2h:20m. But is that perception skewed towards a few outliers, or are films actually getting longer? The answer is, as usual – complicated:

Film Runtimes 1990-2013

Two interesting patterns emerge:

First, the top-30 grossing movies are consistently longer than the top-100. That’s super interesting, but we should not rush to the conclusion that longer equals more successful. The correlation between runtime and domestic gross is actually very low for the top-100 (r-squared < 0.1) – a fairly odd result, which may imply that these blockbusters are unnecessarily long.

Second, for 22 years – between 1990 and 2011 – movies’ runtimes were pretty much steady (less than 6% fluctuations) for both the top-100 and the top-30. But then in 2012 and 2013 – the top-30 shot up by about 15%, while the top-100 remained steady. Is this the beginning of a trend? We’ll have to wait a couple of years to see about that.

But let’s look at a histogram of all these 2,400 films:

Film Runtime Histogram 1990-2013

We see that this is not a pure bell-shaped Gaussian distribution: it is skewed to the right (the average is beyond the peak). This means that when films get longer – they can get really long.

But in any case, one thing is true: if you are like most Americans and primarily pay for tickets for the top-30 biggest blockbusters – you are not wrong if you feel these films are getting longer. But if you mix smaller films into your cinematic portfolio, that is not the case.

So why are blockbusters getting longer? Perhaps the producers want to justify the ticket price by providing the viewer with a long evening of cinematic entertainment – maybe even too long. Or perhaps the huge upfront costs of setting up a blockbuster production make the marginal cost of an extra 30 minutes of film negligible? Or perhaps it’s the competition from premium television’s longer and more complex story arcs?

* All the raw runtime data is from (100 films for each year; 2,400 films total. Yes – a Python script did that).

– Like this post? Follow me on twitter: @stavjdavis and/or like the Facebook page



This is the third part of the International Box Office Correlations series. And it is worth the read, because this time we go beyond the scope of Hollywood, and plant the seeds for world peace, as well as explain geo-political turmoil! In the first part, we looked at grosses in the US vs. the rest of the world combined. In the second part, we looked at the US vs. every country separately. This time, we look at every country vs. every country, in what has turned out to be a giant colorful correlations-matrix that looks like a square Caprese salad. (The image is pretty heavy, so give it a fews seconds. It is followed by two half-matrices, so you can actually read the numbers).

International Box Office Correlations Matrix

International Box Office Correlation SubMatrix 1

International Box Office Correlation SubMatrix2

How to use this?

Just pick a country on the vertical list and cross it with another one on the horizontal list. Darker green means a stronger correlation between the two countries. Darker red is a weaker correlation. White is somewhere in the middle.

A few observations

1. China, Nigeria and Japan are red all across. This means they are very individualistic in their acceptance of American films (be it due to local taste, language, local cinematic product, economics, distribution deals, politics, regulation, etc.). These three countries are not only different than the US – they are different than every other country – and that is not trivial.

2. Other “red” countries are India, Indonesia, South Korea, Uruguay, Thailand and Malaysia. But at least they have a few green cells in their arrays.

3. However, as opposed to the examples of Nigeria-China or China-Japan’s low correlations, in the case of India and Indonesia, they actually have a very high correlation between themselves, which means they probably share the same reasons for being so different from the rest of the world. Other examples of countries that probably share similar factors are: South Korea, Indonesia and Malaysia (red across the board, except among themselves).

4. The “greenest”, or “most-agreeable” countries are: Norway, Italy, Hungary, France, Russia, Germany, Australia, Mexico, Brazil, Turkey and South-Africa. This is a surprising list, because some of these countries are commonly perceived as having a very independent culture. According to this table, perhaps these should be spearheading international peace talks, as they seem to be the most relatable by the rest of the world :- )

5. Speaking of world peace: Lebanon’s second highest correlation is with Israel (0.79). Perhaps this will spark peace in the Middle East. The Netherlands and Germany are also closer than their soccer fans would like to admit (0.92). Turkey and Greece are as well (0.76) – don’t forget to thank HollyQuant when Cyprus is soon united.

6. In light of the recent events in Kiev, this seems relevant: the highest value in the matrix is Russia-Ukraine: 0.96. That is an extremely high value, even if the connection is obvious. This should probably trouble the pro-west protesters as it shows Russia’s control over the country through the CIS economic agreements. And all kidding aside, the recent cost of human life in the Ukraine is a real tragedy and I don’t really mean for this post to be anything more than just one more humble perspective on that matter. Even though this high correlation may be manipulated into supporting the claim that the Ukraine should remain a de-facto Soviet Russian entity, I believe that in this case, the number is not rooted in cultural similarities, but rather serves to prove the Russian economic stranglehold over the Ukraine – a stranglehold that is finally being cracked by revolution and bloodshed.

7. Other high correlation clusters can be spotted between the South American countries (e.g. Argentina-Chile (0.94), Argentina-Colombia (0.92), Bolivia-Peru (0.94), Colombia-Bolivia (0.93), Bolivia-Ecuador (0.93), to name a few); the Northern-European cluster (e.g. Finland-Denmark (0.92), Germany-Finland (0.91), Netherlands-Denmark (0.91), Netherlands-Germany (0.92), Norway-Denmark (0.92), Norway-Germany (0.94), Norway-Netherlands (0.92), Sweden-Finland (0.91), etc.); other European pairs (such as Belgium-France (0.91)  Austria-Germany (0.94), Russia-Ukraine (0.96) and others); an East-Asian cluster (Singapore-Malaysia (0.91), Philippines-Singapore (0.91), Thailand-Malaysia (0.91); a weaker English-speaking commonwealth cluster that includes Australia, New-Zealand, the UK and South Africa.

8. Another important observation is that all numeric values are positive. This means that while some of the countries are extremely weakly correlated with others, none of them are negatively correlated (+1 means perfect correlation; -1 means perfect opposite correlation; zero means no correlation at all). This makes sense but is interesting nevertheless: none of the countries have an opposite behavior than any one of the other countries, when it comes to American movies.

9. Have more observations? Notice any surprisingly non-green pairs? Or non-red pairs? Please share!

How was this done?

A python script was used to collect domestic and foreign box office data for every country available on – for the top 100 domestic films in 2011, 2012 and 2013 (300 films total). The table here is the correlations matrix of all those countries, based on the 300 film titles. A higher number means a stronger correlation. Theoretically, the numbers can be anywhere between -1 and +1 (-1 means perfect opposite correlation; +1 means perfect correlation; zero means no correlation at all). However, all the values in this table happen to be positive.

– Like this post? Follow me on twitter: @stavjdavis and/or like the Facebook page


In the last post we took a look at the correlation between box office performance in the US and in all foreign territories combined. This time we take a look at the relationship between US box office and each territory separately. In other words: which countries have the most similar box office patterns to those of the US?

US vs Specific Countries Box Office Correlations

What does this mean?

A strong linear relationship, represented by a high r^2 value means that better box office performance in the US is strongly linked to better performance in a specific territory. Correlation is not causality, so we can NOT officially conclude that Australians, for example, just mimic Americans, but we can say that if a movie is released first in the US and does well, it is more likely to do well in Australia than it is in Slovenia, for example.

“Well, duh”, you may say – it’s no surprise that English speaking countries top the list. But where it gets interesting is when we notice that the United Arab Emirates and Egypt are way up there as well. Or that Eastern European countries are not necessarily infatuated with American hype. Or that the South American countries are widely spread across the list and are not as homogenous as stereotypically assumed. Or that China is way down at the bottom, contrary to some people’s impression that the Chinese have nothing to do but wait for Will Smith.

This list does not refer to the relative size of each market, but rather to the “ease of approach” to that market, so you may still prefer to target the Chinese market due to its sheer size, but I think it’s important to factor this gauge of local openness towards american cinematic product into the equation. Perhaps it will make some marketing team target Brazil before it targets Russia. Or maybe even Egypt before Japan (Hmm… Probably not, but still an interesting thought).

The list is of course affected by many variables: links in marketing budgets; varying cultural taste; economies of scale and scope in distribution; local industry competition; international politics and finance; pro- and anti-american sentiments; and many other potential reasons. Please contact me with any ideas about this!

How was this done?

A python script was used to collect domestic and foreign box office data for every country available on – for the top 100 domestic films in 2011, 2012 and 2013 (300 films total). This plot shows the r-squared values, representing the strength of the linear relationship between box office grosses in each territory and box office grosses in the US, based on the 300 titles.

– Like this post? Follow me on twitter: @stavjdavis and/or like the Facebook page


In the new HollyQuant spirit of “more python code, less talk”, here are today’s findings:

Box Office Totals: U.S. vs. Foreign

How was this done?

A python script was created to collect domestic and foreign box office data for every country available on – for the top 100 domestic films in 2011, 2012 and 2013 (300 films total). This plot shows the non-U.S. total (all countries combined) for each of the 300 titles and their fairly strong correlation (r^2=0.71) with domestic performance. A breakdown by country will be discussed in the next HollyQuant post (very interesting findings there).

What does this mean?

This means that while U.S. box office may be losing ground in terms of global market share (see older post on that), it’s still a good indicator of international box office performance. While we keep in mind that correlation is not causality (therefore, we should not conclude that U.S. buzz is a direct factor of international success), what we can say is that if the film was made by serious people who properly marketed it in the US, it has very good financial prospects abroad. Every gross dollar earned domestically translates into an additional 1.57 dollars in total foreign grosses. Of course there are always outliers (Disney’s ratio is probably better than Lionsgate’s for now), but if you were to blindly choose a bunch of random titles, this would be a good estimate.

The next post will discuss the super interesting finding regarding the correlations of domestic performance and performance in each specific country.

– Like this post? Follow me on twitter: @stavjdavis and/or like the Facebook page


From my experience, when Hollywood executives hear the word “drama”, they tend to suddenly remember a prior engagement, politely end the conversation, exit the room and start running until their distant cries of hysteria blend into the massive exodus of other industry professionals who caught word of the project and are now fleeing in any vector that is pointing away from the mentioned production.

It seems that this notion is one of the most violently engraved paradigms of the industy’s psyche. It’s as if the rule of thumb is: “unless it has a real shot at major awards, stay away from dramas”. Comedies are cool (we’ll see about that), horror is awesome (not), action is just dandy (sure, if your underwear is made of pure gold). But dramas seem to be lepars. And no one I talk to can really pin point why. So I set out to investigate the numbers.

The first step is also the simplest one: finding the Return On Investment (ROI) of different genres. This was done by aggregating all the films released on more than 500 screens, between 2008-2012 (kudos to our new data mining tool at Great Road Capital), that have production budget data publically available on – and averaging the ratio of domestic box-office-to-budget, for every genre. Here is what it looks like:

Domestic ROI by Genre

This is a great example of why I don’t like averages and means, but we’ll get to that in a bit.

First, we can see that there isn’t a significant difference between “heavy award contenders” and just regular dramas[1]. In fact it looks like “Pure Dramas” do slightly better than the much larger group of drama sub genres (which may include films that are really action/thriller/fantasy films with a small drama element). Second, we can see that while horror is way up there, dramas actually do better business, on average, than action movies or, surprisingly, comedies. More about that at the end of this post.

So that didn’t quite explain the hatred towards dramas.

The next thing on the agenda is to try to define risk. Anyone with a basic understanding of statistics knows that averages can be very misleading, if they are not coupled with some kind of estimate of the spread around that average/mean. One of the first things we ask at GRC when presented with a seemingly compelling “average argument” to support a pitch for a project, is “what is the standard deviation?” and “what is the size of your sample?” (and of course – “did you cherry pick that sample?”). So in order to get a better look at the genre comparison, the risk for that same group of films was defined as the normalized ratio of the standard deviation to average of the domestic box office (don’t worry about the sample size – it’s at least 50 films[2] for each category and I promise I have no agenda here, so cherry-picking did not take place). Here are the results:

Domestic Risk by Genre

Now we are starting to see why producers and development execs have a bad hunch about dramas: while they may do well on average, they are a wild card, especially compared to comedies. If I could invest in a thousand movies – having all of them be dramas could be a good idea, but if I have to randomly pick one investment, dramas are just too unexpected. Horror is less volatile than I expected, but don’t forget that this study only includes the movies that got a 500+ screen release, and not all of the trashy low budget productions that barely make it to the production finish line and end up on 3 midnight screens in New Mexico at best. Action movies are even more stable than comedies, but that’s a reward that only the big players have the resources to tap into.

The third aspect that seemed very relevant is each genre’s dependency on budget. This was calculated by normalizing the r-squared value of the domestic box office vs. production budget spread plot. Here are the results of that:

Domestic Dependency on Budget by Genre

Here we see that big studios and their action flicks are the big winners: every dollar invested in the budget is far more likely to yield an extra dollar at the box office in this genre than in all the others. Comedies and dramas have a budget-to-box office relationship that is about half as strong and horror is way behind: a more expensive horror production does not translate into better financial performance (which reduces the advantage of big players in this genre – more on this in a bit).

So, to summarize, here are all three plots together:

Domestic ROI Risk and Dependency on Budget by Genre

What this all means:

If you have a huge amount of capital, like studios do, it makes sense to go for action mega-movies, because while their relative return is lower, their absolute return can be higher and their risk is relatively low, plus you know that every dollar you put into the production is probably going to have an impact on revenues.

While lower risk is the privilege of the rich, if you are a medium sized independent production company, comedies are a nice option[3]: a relatively low risk (yes, I know that’s a kind of oxymoron in this industry), coupled with un-spectacular-yet-fairly-solid returns. A medium dependency on budget means that you have a chance of making it work even if you don’t have a giant pool of cash, while still keeping the barriers to entry high enough so the landscape is not too crowded (which is the biggest problem for horror productions). This is, of course, not to say that comedies are always winners, but they are less of a wild card.

Moving on to the low budget production houses and their love of horror: even after the exclusion of “Paranormal Activity” from the sample[4], there is a very low correlation between budget and performance – which is good if you, well, don’t have much of a budget. However, this translates into low barriers to entry, which means that everyone is trying to do horror, resulting in performance that is extremely unpredictable (and that doesn’t even include the ones who never made it to the finish line). When we get pitched horror scripts, we need to see why a specific one is unique: it’s not enough to rely on the memory of a few recent Cinderella stories, because for every meteor horror success story, there are so many crashes-and-burns. It’s very hard to make horror that is both good and successful, as proven by these numbers.

And finally, dramas: now that we have a more detailed picture of film performance elements (on top of the simplistic misleading average performance argument), you can see why dramas are the pariahs of the film industry: while they have a reasonable relative return, they have an insane risk level, regardless of their award-worthiness or purity of genre. In addition, they have a medium correlation between budget and performance, meaning that only medium-large players can effectively exploit those seemingly attractive returns.


[1] Note that while the many sub genres of drama may include things like “fantasy drama” or “action drama”, the category of “Pure Drama” includes only films categorized by as “drama”. When evaluating that list, the vast majority of those were accurately categorized as such hardcore dramas, with only very few odd exceptions that were omitted.

[2] Sample sizes (number of films with available production budget and domestic box office data): “All Drama Sub Genres” – 124; “Dramas Not Including Best Pic Noms” – 107; “Pure Drama” – 97; Comedies – 219; “Horror” – 61; “Action” – 80; “All Movies” – 718.

[3] This recommendation is obviously a bit superficial, as we can dig into many additional important details, such as the international-vs-domestic appeal of this genre, which is crucial for independent productions as international pre-sales are the key to debt financing of the budget, while domestic appeal is the key to actual revenues (in most cases).

[4] For the horror genre, the film “Paranormal Activity” was left out of the list, since it’s performance was so far out of range (it had an extremely abnormal behavior: 15,000x ROI) that it managed to skew the results, even though the sample size was large.

A few general notes and caveats:

– These results refer only to domestic box office. The picture might be different for international box office.

– The list of movies evaluated here pools together studio productions and independent productions, which have very different financing and development processes, so the conclusions might vary depending on specific project conditions.

– Of course, production budgets are different than total budgets (which include marketing and P&A budgets – that have been proven in this blog to have a significant impact on performance), but given the large sample sizes for each genre, these results should still provide a pretty good picture of the general performance trends.

– Like this post? Follow me on twitter: @stavjdavis and/or like the Facebook page


This post was originally published here a couple of months ago and included a plea for help in making the data more approachable  via better infographics than those my ugly hands can produce with Excel/Matlab. RJ Andrews who runs came to the rescue and together we collaborated to create this first class infographic. He did an amazing job and I hope we will work together again!

We took a look at Lionsgate – often called a “mini-major studio” or “the sixth of the Big Five”. How did this happen? How could a company that was founded in 1997 by a banker from Vancouver take on the film industry under its own terms and without the backing of a major media conglomerate like the other big ones? Many voices say they got lucky by hitting gold with The Hunger Games right after acquiring Summit and the rights to the Twilight Saga. But that’s just the tip of the iceberg, in what turns out to be a very deliberate, aggressive and strategic expansion effort over the past 15 years.

The company got to where it is primarily by making bold, brave, risky and sometimes stupid moves. Don’t get me wrong – they made some films that were utter nonsense, but they also took a risk on American Psycho after Disney got spooked by its uneasy premise. Or Lolita, which had a hard time finding a home in the late 90’s. Or Kevin Smith’s Dogma, which required some guts to believe in before it became a cult hit.

And then there are the strategic investments in streaming video as early as 2000(!) and its focus on the growing Latino film market.

At the core of all this are the strategic acquisitions that Lionsgate has been putting together like crazy over the past decade, focusing on expansion of its title library and distribution network – domestic and international, as well as its independent production capabilities, strategic partnerships and expansion into TV. Independent film production companies should study this if they want to break away from the pack.

Here is our beautiful infographic. A huge thanks again to RJ Andrews for this – it looks beautiful, if I may say so myself.


* For the backstory of the infographic – check out RJ’s post on his blog.

* Like this post? Follow me on twitter: @stavjdavis and/or like the Facebook page