There are really two claims tangled up in that reputation.

1. The first is that the list finds talent, i.e. it is very good at picking out the scripts and writers who matter.

2. The second is that it makes careers, as in, simply being on the Black List actively lifts a writer who would otherwise have stalled.

Those two things are not the same, so let’s test each in turn.

First, just how well do these scripts do?

Start with the scripts themselves. Of the 1,901 screenplays listed between 2005 and 2025, 576 have been made into films, and together they have grossed over $31bn worldwide at the box office.

That includes 103 films that grossed more than $100m each. Four Black Listed scripts have won Best Picture (Slumdog Millionaire, The King’s Speech, Argo and Spotlight), and a total of 74 have received Oscar nominations.

But raw totals can flatter. The fairer test is how these films compare to everything else released in the same era. So I plotted every Black List film against the wider field of films from 2005 to 2025, on two axes that matter: what audiences thought of them, and what they made.

The Black Listed projects are represented by the orange dots, and they are not randomly scattered. I have added dotted lines at the median film, meaning that 25% of any random group of films will fall in the top-right quadrant of films that are both well-liked and commercially successful.

For Black List films, 55% land there, more than double that base rate.

Three-quarters out-gross the average film and nearly two-thirds out-rate it. The typical Black List film earns around $23m worldwide and scores 6.5 on IMDb, against 6.3 for the field. These are, on the whole, good films that make money. As a piece of talent-spotting, that is a very strong hit rate.

I am not the first to find this. In 2019, researchers at Harvard Business School ran their own numbers on the annual Black List and reached strikingly similar conclusions. They found that listed scripts were about twice as likely to be produced, and the resulting films earned roughly 90% more at the box office than comparable titles made on the same budget.

Does the Black List spot talent?

Let’s move from the produced movies to the writers behind them. If the list really has an eye for identifying talent, then the writers it picks should be on the rise.

To check, I used my own measure of fame (based on a mix of public sources) to track our writers’ visibility around the year they first appeared on the list.

Before the list, these writers are in a much lower position, and that position is static. Once they get listed, the line climbs and keeps climbing, roughly doubling within five years.

So the answer to the first question is clear. The Black List spots talent extremely well. Its scripts succeed, and its writers go on to careers that visibly take off.

Does the Black List make careers?

The climbing visibility line we’ve just seen certainly does look like a career launch.

But we need to consider causation here. The reason the writers have ended up on the Black List is that they’ve written something that has got some serious Hollywood folks talking about it. They just wrote something brilliant, meaning that the listing lands, almost by definition, at a moment when their career was already heating up.

How much of the rise is the list, and how much would have happened anyway?

One way to answer that is to compare these writers to others who were in exactly the same position but did not make the list.

So I built that comparison. For each Black List writer, I found a matched control: a writer with a released film in the same year and a similar number of prior credits who had never appeared on the list. Then I ran the identical before-and-after analysis on both groups.

Black List writers roughly doubled their output in the five years after listing, a factor of 2.15. And the comparable writers who never made the list managed a factor of 1.96 over the same stretch.

It’s fair to say that some of the “Black List bump” is what happens to any writer you catch at a hot moment, but not all of it. Black Listed writers sit consistently above their matched twins, outproducing them by roughly 18%.

So being on The Black List provides a modest bump to an already on-the-up career.

It is not just writers it launches

Before we finish, here are two interesting tidbits I found along the way. While the Black List bills itself purely as a writers’ showcase, it is also providing fuel to directors’ careers.

More than a quarter of Black List writers have directed a feature film, and on a like-for-like measure they do so at over twice the rate of writers in general (35% vs 15%).

A full quarter of the Black List films that got made were directed by one of the script’s own writers, a group that includes Paul Thomas Anderson, Damien Chazelle, Emerald Fennell and Taika Waititi.

How quickly does it launch people?

The second data point I found concerns the delay between the script’s appearance on the Black List and its reaching cinema screens.

For the scripts that get made, the median time from listing to release is three years, and more than three-quarters arrive within five. The chances of a sale peak in years two and three and then fade away, so a script’s best window is early.

(That's looking at the share of films that are eventually made. Only about one script in three has ever been made, and most of those arrived within five years of listing).

So the launch, where it happens, is reasonably quick. It just does not happen for most scripts.

The Black List is a superb spotter of talent, a modest maker of careers, and, for the majority of the screenplays it celebrates, a beautiful credential that never becomes a film.

Reaction from Franklin Leonard

Before publishing a study, I like to reach out to one or two people to see if I’ve missed anything obvious and to provide context. For today’s article, I chatted with the man behind the Black List himself, Franklin Leonard.

He’s a math whizz, so rather than being affronted that I turned his life's work into a few charts, he really engaged with the findings. He was also careful not to overclaim on what The Black List is and is not:

I try to be very careful about what we claim, because the truth is already strong enough, as your analysis shows. I’ve never believed that the Black List makes careers. A writer’s career takes off because they wrote something great, something good enough that the right people couldn’t ignore it.

What the Black List does is put a very bright spotlight on that script so that the people who should read it actually do. And once they’ve read it, they get involved with that script or come back to the writer for the next thing, and the next thing, and so on and so forth.

Ultimately, we’re a catalyst, and a catalyst isn’t a small thing, especially in a system so broken that an almost laughably rudimentary data study can consistently beat the market.

And it’s still an incomplete solution to the deeper discovery problem. The annual list can only do what it does for scripts already in circulation, the ones that are already in the right rooms.

That’s actually why we built the website: To be the same catalyst for writers who might not be in the room yet at all. But the principle’s the same either way: If the work is good enough, it should have opportunities commensurate with its merits, regardless of who wrote it, where they live, or who they happen to know.

I’m very grateful to Franklin for his good nature and time.

Notes

I looked at every annual Black List from 2005 to 2025. That’s 1,901 scripts, 2,222 writing credits, 1,682 writers matched to a verified IMDb profile.

On the scatter plot, each point is a commercially tracked film released from 2005 to 2025 with a worldwide box office figure, an IMDb rating, and more than 1,000 votes (to exclude obscure titles). Box office is on a log scale. Black List films are those whose listed script reached the screen.

When creating the career control groups, each Black List writer was matched to one non-Black List writer with a released feature in the same year and a similar number of prior credits, drawn from the full IMDb writer population. Both groups are measured using the same before-and-after method, which ensures a fair comparison.

This research shows correlation, not proof of cause. The matched comparison strongly suggests that most of the career rise is due to selection and momentum rather than the list itself, with a smaller genuine lift on top. My measure of fame measures attention, not quality, and reacts to events such as releases, awards, deaths, and scandals.

The Harvard Business School study referenced above is by Hong Luo and colleagues. The accessible write-up is at HBS Working Knowledge, with the academic record here.

1. “We’re moments away from uploading a script and making movies!” (This is usually followed by the person trying to get me to look at a demo of their “ground-breaking” vibe-coded AI-to-storyboard start-up).

2. “It’s all hype and will burst any second, returning moviemaking to how it’s always been” (These folk also complain that GenZ don’t understand customer service and the generation below that can’t read).

Between these two outlier options sit the rest of us - curious, confused, and tired of having to work out what’s bullshit and what’s best practice.

I was recently asked by a highly experienced Hollywood VFX veteran what the lay of the land actually is today, so I thought I would dive into the facts and find out how things stand as of July 2026.

I sought out every report of a named film or show where AI genuinely did some of the work. I also looked at what the myriad of “AI-powered” tools, apps, and services are promising (taking special attention to which had real-world case studies and which did not).

My main takeaway is that the AI solutions that are working in professional settings today are mostly the boring kind. They are not the ones which make for the most exciting/scary headlines, but the ones which save time and money.

The tools shouting loudest are often the ones doing the least, whereas some of the most useful ones don’t (or maybe don’t need to) do much public-facing marketing.

Below is an illustrative chart of how I see things:

Let’s dive into the details.

Where AI actually works in visual effects

Let’s start with what some might call the thin edge of the wedge. This is where AI is fully integrated and accepted by all. VFX artists use AI today to have computers perform the most tedious manual tasks, and to do so so much faster than humans can do alone.

A good example is on Dune: Part Two, where visual effects supervisor Paul Lambert needed to turn the Fremen’s eyes spice-blue across around a thousand shots. Instead of painting each one by hand, the team trained Foundry’s Nuke tool CopyCat (a small neural network you teach on a few frames) to do it. Foundry reported that 40% of those eye shots needed no manual touch-up at all.

Other examples include:

• For Furiosa: A Mad Max Saga, the Australian house Rising Sun Pictures used its REVIZE machine-learning system to blend Anya Taylor-Joy with a younger performer across about 150 shots.

• On Kingdom of the Planet of the Apes, Wētā FX ran a facial deep-learning system across more than 1,500 shots.

• The Toronto company MARZ sells a beauty-fix tool called Vanity AI and says it has run on more than 27 productions.

AI has already come for people’s jobs, but it’s starting with the jobs no one wanted in the first place. The artist remains in charge, directing the robots toward a single, repetitive task.

This is clearly the safest ground for VFX companies to talk about AI use without causing a backlash. That’s because:

1. It is replacing work no one wants to do.

2. It does so without using anyone else’s art. The companies train on their own images and haven’t built a tool others can use to create new art.

3. We can all picture what’s happening, on a conceptual level - it’s not that scary.

Not all AI is AI, actually

Next up, we have “AI de-ageing”. We’ve been hearing about this for years, and the truth is, a lot of it wasn’t really AI. Well, not the Generative AI that we think of today, where you type in a prompt and get out a fully generated image.

In recent years, we’ve seen a de-aged Robert De Niro in The Irishman and a young Will Smith in Gemini Man. In the case of the former, ILM used a markerless capture rig and extensive traditional computer graphics, and for the latter, Wētā built a fully digital human. These are CGs made by hundreds of people, and where machine learning was used, it was more of a reference aid.

The visual effects supervisor on Gemini Man, Guy Williams, described it like this:

We didn’t even think of it truly as de-ageing; we thought of it as a complete creature replacement

A film that came closer to “AI de-ageing” was Robert Zemeckis’s 2024 film Here, which used a neural network rather than CG to de-age Tom Hanks and Robin Wright live on set. The company behind it, Metaphysic, ran the face replacement as the cameras rolled (albeit that’s not to say that what we saw in the final film was the real-time output).

So when you read that a film “used AI to de-age” a star, the question worth asking is whether they mean “Here,” or the other 90%, which is that skilled humans do the work.

Has generative AI reached professional screens yet?

But that’s not to say that Generative AI isn’t being used. An early example of such a project is El Eternauta, an Argentine Netflix series that used generative AI for a building-collapse shot. Netflix co-chief executive Ted Sarandos told investors a year ago that it was about ten times faster than the traditional method and that it was:

The very first GenAI final footage to appear on screen in a Netflix original.

So it was a year ago that just one AI-generated sequence making it to air earned a press release and headlines. Things are not as far along as some will have you believe.

Read more

The article critiques the work of the USC Annenberg Inclusion Initiative, which studies the gender of top-grossing directors, the festival circuit, animation, visual effects, and more. Hollywood Gadfly raised several objections to their work and methodology, including whether it is fair to treat the gap between the share of women directing at Sundance and the share directing top-grossing films as evidence of a broken pipeline.

The chart below is one such example of how Annenberg makes the case, highlighting the Sundance figure at 63.6% (in 2026) and the top-grossing directors at just 6.6% (2007-25).

Leaving aside that these figures cover very different periods, this raises the question:

Is there a pipeline from being selected in the US Dramatic Competition at Sundance to top flight directing?

I studied the career paths of 1,765 directors, all of whom have directed a top-grossing movie since 1980. I looked at the professional work they had done before they directed a movie that appeared on the ‘top 100 US-grossing’ chart that year. More detailed information about my methodology is at the end of the article.

Where do directors come from?

I narrowed in on each director’s first appearance in the annual top 100 grossing films at the US box office.

I looked at their past work, tagging it with several types of activity, which also means that one person could have multiple tags, since a few people just stick to one thing.

Three-quarters of them had worked in television prior to making their first top-grossing movie. Almost the same number had been credited as writers on a TV show or movie, and just over half had directed a feature film that was not one of the 100 highest-grossing movies in the year it came out.

Given how big writing is, I thought I’d split that out. Writing movie screenplays was the biggest type of writing gig that top-grossing directors did before reaching the top 100.

So does the Sundance discovery pipeline hold up?

The core idea is that people who go on to become top directors are discovered on the festival circuit. Specifically, being in the dramatic competition at Sundance is a meaningful springboard to a studio career.

I tracked eighteen of the discovery markers - the competition strands at Sundance, Cannes, Berlin, Venice, Toronto, SXSW and Tribeca, the first-feature and new-director awards, and the celebrated short-film prizes, and I followed every director who hit each one forward to see how many later directed a top-grossing film.

16% of directors nominated at the Spirit Best First Feature award went on to top-grossing status. That’s slightly above Sundance (14%), an Oscar nomination for best animated short (11%), and the Discovery strand at TIFF (8%).

On the chart below, I have put most festivals in blue but highlighted the main competitions at Cannes, Berlin, and Venice in orange, as one could argue these are not really seen as ‘discovery’ competitions.

So Sundance is the second-best choice for comparison, but that’s from within a batch of pretty weak signals.

If you present a pipeline that is just “Sundance to top-grossing”, then you would be looking at a small minority of the people who actually make it, while the meaningful markers are having worked in writing and television.

What is the churn rate for top directors?

Another thing to consider when assessing the efficacy of using a pipeline model is the churn rate. i.e., how many people each year get to direct a top-grossing film for the first time.

Interestingly, this has been fairly static for three decades, but then it has started to rise in the past ten years. Until the mid 2010, around a third of directors of top-grossing films had not previously had a title in the top 100.

2020 was a freak year, thanks to Covid, in which 71% of directors in the ‘Top 100 Club’ were new (77 of 109 directors). This is affecting the five-year rolling average, but that’s not the sole cause of the recent rise. There was also a real, sustained step-up beginning around 2016, from a long-run norm near 33% to a new normal around 40% to 45%

The feeders themselves have shifted, too. Writing and television have always led, and both have grown more common among recent arrivals, with editing and acting rising alongside them.

Some of this is down to careers becoming more multi-stranded, and it’s possible that some of it reflects the fact that recent careers are simply better documented than those that began in the 1980s. Either way, the ordering is steady throughout, with writing and television on top the whole time.

Earlier this month, I wrote about how the average age of directors has risen a lot in recent years, so you may want to check out that article for context, too.

Do the routes treat women and men the same?

Now that we have a more detailed and nuanced picture of how people become top-grossing directors, we can split the cohorts by gender and assess the extent to which gender is a factor.

Let’s start with how many women reach the top 100 at all. The share is a long, uneven story rather than a steady climb. Women were close to one in ten of new top-grossing directors in the early 1990s, slipped back through the 2000s, and only in the 2020s reached around 18%.

I pooled all the early-career routes, festival competitions, first-feature awards, and short-film prizes and followed only those who had attained them by 2012, so everyone has had more than a decade to convert.

One in ten male directors from that pool went on to be a top-grossing director, compared with just 4% of female directors.

On the main feeders, women and men look broadly alike. Both come overwhelmingly from writing and television. If anything, women lean into those two even more than men (85% of women had worked in television first, against 75% of men).

The biggest skew towards men was among cinematographers stepping up to directing, which was a small route in and picking from a field that has historically been massively male.

The festival discovery route is where we see a big difference in outcomes. Men out-convert women at the Sundance dramatic competition, at Cannes, at Venice, at Toronto's discovery strand, and at the Oscar short categories.

The direction is the same almost everywhere, and it holds across four decades, so it is not a matter of women having arrived too recently to have converted yet.

The women who do come through these routes are exactly the names you would expect - the directors who went on to make some of the most admired films of recent years. So the route does work for some women. It simply works for a far smaller share of them than of the men who start in the same place.

The other biggest difference between the genders is getting a second crack at the big director’s chair. Reaching the top 100 once is hard for everybody, and most directors never do it twice. But the rate at which they get a second turn differs.

Among directors who arrived by 2018, with plenty of time to make another film, 53% of the men went on to a second top-grossing film. For women, the figure is 40%.

Over a whole career, men in this group average about 2.8 top-grossing films each, women just 1.9.

What does all this tell us about directing careers?

We’ve gone through a lot in this article, so I thought I’d put a handy list of things we’ve learned along the way:

• Getting the job as a director on a movie that ends up in the top 100 highest-grossing movies of its year most commonly comes to men in their early forties, after years of other work.

• Women and men enter the top 100 through the same main doors, those of writing and television.

• A third of the seats turn over each year, while a stable group of returning names holds the rest.

• The festival and award routes contribute a real but minor share, and none of them comes close to being the pipeline.

• The gap between genders emerges at the discovery routes, which carry men through at more than twice the rate, and after the first film, where men are markedly more likely to get a second.

• Some directors do go from being in the official selection at Sundance to becoming a top-level director, but not many. It’s one of the biggest routes within the festival discovery framing, but a poor one overall.

Notes

I’m very grateful to both the pseudonymous Hollywood Gadfly and to Dr Katherine Pieper, Program Director of the Annenberg Inclusion Initiative, both of whom generously talked this through with me over Zoom.

For today’s research, I counted a “top-grossing film” as one of the 100 highest-grossing releases at the US box office in its initial year of release. A “top-grossing director” is anyone who directed such a film, and a “first-time” top-grossing director is one reaching that group for the first time. Co-directors are each counted. My study covers first appearances from 1980 to 2025, 1,765 directors in all. This top-100 framing follows the USC Annenberg Inclusion Initiative’s annual analysis of the director’s chair, which I rebuilt in order to follow each director’s career forwards.

Box office data came from The Numbers, Box Office Mojo, and Wikipedia. Directors, prior roles, release years and dates of birth are from Wikidata, OMDb, IMDb, The Numbers, Opus and other data sources. The festival and award records, including the dramatic competitions and first-feature and short-film prizes, are from my awards database covering 1978 to 2026; for each route I took the directors of every film in it, not only the prize winners.

When I mention film in competition at Sundance in today’s piece, I am specifically referring to films in the Sundance US Dramatic Competition. Sundance has a number of selections beyond this.

A minor footnote on the Sundance figures is whether the festival film itself reaches the top 100, or if the director gets there later with a subsequent film. It is almost always the latter, with only about one in seven cracking the top 100 with the film they premiered at Sundance. That share is inflated thanks to the strange pandemic years, when the box-office bar fell low enough for small Sundance titles like Minari and Zola to slip in.

For each route, a director “converts” if they directed a top-grossing film after the route year, so the route plausibly preceded rather than followed their establishment. Measuring it at any time gives higher figures for the main competitions, because established directors keep playing them, which is why the directional measure is used and those routes are flagged. Eighteen routes were tested in all; the gender comparison for them is shown for people who reached the route by 2012, with thirteen or more years to convert, and the gap holds in that long-runway pool (women 4%, men 10%). Per-route female counts are small for some routes, so the pooled figure is the reliable one.

Prior roles come from credited work, so uncredited jobs, foreign-language work and very early-career roles can be under-recorded, and this is more true the further back you go. Recent arrivals show more prior credits of every kind than those from the 1980s, which is partly a real change and partly better record-keeping, so the prior-role shares are best read as a floor, especially for older directors, while the relative ordering, with writing and television on top, is more reliable than any single figure.

A director’s gender was applied across their whole career, using their present-day identity. This is a very minor point from a purely data perspective, which is why it is in the Notes rather than flagged in the main article. However, I know it will animate some people on either end of a topic where feelings run high.

It’s relevant here as both Wachowskis identify today as transgender women, using the pronouns she/her, but were seen as men when they co-directed their first top 100 film The Matrix in 1999.

I cannot profess to have any special insight into the best way to tackle this. It’s a deep topic that many people feel aggrieved by. I would note that the comments section of this article is hardly the right forum to debate something as big as ‘What is a woman?’ No one is going to solve anything, nor change anyone’s mind.

Whichever perspective you take, this decision accounts for just 1.9% of the directing credits for the year 1999 and 0.27% of all directing credits I studied.

I genuinely want to hear from everyone in the comments. Feel free to share your thoughts, ask me anything, challenge me, etc. Just be nice to each other. I won’t delete comments I disagree with, but I will remove nasty ones.

Further reading

If you would like to read more about the statistics behind gender in film, then you may enjoy the following: (ok, bad choice of words - you may wish to read the following):

• A major new study into gender inequality in the UK film industry (2016)

• UK films with public funding hire more women (2016?)

• Major new study into gender inequality among UK film and TV writers (2018)

• Are movies dominated by a small number of old male composers? (2018)

• Are women less likely to direct a second movie than men? (2019)

• What percentage of film producers are women? (2019?)

• How often are women hired in key film departments? (2019)

• How well are women represented among short filmmakers? (2020)

• Do male and female critics judge films differently? (2021)

• What percentage of movies are written by women over 40? (2024)

• Is Jason Blum right that there is a shortage of female horror directors? (2025)

• How do women fare in gender-neutral Oscar categories? (2025)

In the past couple of months, two films made silly amounts of money. Obsession was made for about $750,000 and has already grossed $297 million worldwide, and Backrooms earned $257 million worldwide on a $10 million budget.

This is the dream every film investor has when funding a movie - a relatively small bet pays off hugely.

Despite seeming similar on the surface, these two movies took a very different approach to financial risk. Obsession was sold to Focus Features for around $15 million at last year’s Toronto Film Festival, eight months before release. Conversely, A24 financed Backrooms itself (with Chernin Entertainment) and kept both the risk and the upside.

Taken together, these films reveal a lot about how the film industry treats risk, money and approaches to business.

So I thought it was a good time to tackle one of the most common questions I get asked:

How do you make money investing in movies?

There are many answers to this, so today I’m going to focus on how people make a sustainable business investing in movies, given that:

1. The movie industry has massively asymmetrical returns, with ‘the hits’ providing huge returns, while ‘the rest’ delivering big losses with almost no post-release assets to recoup from. (If you buy a house and house prices tank, you still have a house. If you invest in a film and it doesn’t hit, then all you have is your ‘Associate Producer’ credit and maybe a crew jacket).

2. It is very hard to predict whether any single film will be a hit. This is even harder than the industry likes to admit, and provably so, especially given that the key things which can be useful mild indicators of success are rarely known at the financing stage.

3. This means you need a business model that doesn’t rely on luck. You need to find a model that goes beyond picking winners and pays off over time. (Either this or be honest and just enjoy the gambling).

So let’s dig into these things and learn how some people, professionally, generate returns over time.

Only a small number of movies matter

Possibly the single most important idea to grasp about the film business is power laws. This is where a small number of examples take home most of the rewards.

Most things in life cluster around an average. A good example is human height. Almost everyone is near the middle, nobody is fifty feet tall, and the average tells you what to expect from the next person.

Film money does the opposite. A few films tower over the thousands beneath them, and the gap is enormous.

Below are the worldwide box office receipts for 49,834 films (converted to 2026 dollars). As you can see, there is no “typical” film.

Another way to look at the same data is to line up every film from the lowest-earning to the highest-earning and ask what share of the total each slice represents.

The top 1% of films take 31% of the total box office. The top 5% take just over 70% of it.

And this effect has actually been growing in recent decades.

So a tiny number of films carry the whole industry. Which means most films must lose money, and they do. Big time. Statistically, if someone asks you to invest in a movie and you are interested in making money, your answer every time should be a resounding “Hell no!”.

They may be able to show you averages from the past of their type of movie, but in reality, the “average film” is a fiction. There is the tail, and there is everyone else.

Soooo… how do we work out which movies will sit in that vaunted ‘tail’ position so we can back them?

People have tried very hard to predict it

For forty years, economists, marketing professors and computer scientists have tried to crack film prediction, using budgets, stars, scripts, trailers, tweets and, lately, AI.

I have read that whole body of literature and written a companion guide you can download, study-by-study, with what each one found and whether it holds up.

Subscribers can download a free copy of my “Predicting Movie Hits" report at the end of this article, summing up the academic literature on the topic.

Here’s the overview of the picture from academics:

The numbers in brackets are the year of publication. This matters a lot as the film economics have changed dramatically over the past decade, so many of these studies would not be replicable with modern movies.

• The biggest useful single factor in estimating a film’s eventual revenue is the budget. This can be used to explain 70% of revenue, but, sadly, only 22% of profit, because bigger budgets have higher bars to clear. (1999)

• A screenplay carries enough signal to tilt a slate of bets. (2007)

• Other meaningful factors include genre, an Oscar and good reviews. (1983)

• Activity on a film’s Wikipedia page a month before release accounts for roughly 77% of the opening weekend. (2013)

• A film’s stars do not reliably move the return once everything else is accounted for, a result that keeps reappearing across the decades. (1993)

• Web search data and social media buzz add almost nothing once you know the budget and how wide a film opens. Most “signals” are just another way of measuring what the studio has already spent. (both 2010)

• Film critics predict success, but do not cause it. They spot the quality the audience would have found anyway. (1997)

• Many of the most eye-catching results lean on the opening weekend to “predict” the total, which jumps the explainable share from 45% to 97%. To my mind, that is reading a result, not forecasting one. (2000)

• Underneath it all, the maths itself (1999) is brutal. Box office is so heavy-tailed that forecasts have, in the authors’ words, “zero precision”, and even today’s best AI (2025) is at its worst on exactly the smaller films an investor most needs help with.

Outside of academia, there is a cottage industry of services within the film industry, happy to sell you tools as you try to illuminate the unknown. These fall into five broad groups:

• Theatrical forecasting firms predict box office for films already on the release calendar. These include Gower Street Analytics, Boxoffice Pro, The Numbers, Numero, and fellow Substacker Shawn Robbins at Box Office Theory.

• Tracking and testing firms survey audiences before release to gauge awareness and intent, which becomes the trade’s tracking numbers, such as National Research Group, Screen Engine/ASI, MarketCast, The Quorum, Comscore.

• Signal trackers, such as prediction markets (Hollywood Stock Exchange, Kalshi and Polymarket) and firms inferring meaning from crowds, search and social activity (Parrot Analytics, Spectura, Emberos, RecoBee).

• Greenlight predictors have been a holy grail for decades, estimating revenue from just a script or cast package. Sounds straightforward, but many fall prey to the “average film” fallacy we saw above. It’s been a turbulent corner of the sector, with many falling by the wayside (Epagogix and ScriptBook) and some still active (Cinelytic, Largo.ai, StoryFit, Vault AI, CineCockpit).

• Comparable-title report builders provide real (or calculated) revenues of past films. These are an everyday tool of indie financing, despite producers' rampant cherry-picking when using them for ROI calculations. Companies in this space include Nash Information Services, FilmProfit, FilmProposals, Cinelaunch, Slated, Callaia.

These types of services are good at providing aggregate market-level estimates, detailing what’s happened before, and in predicting performance in the weeks leading up to release.

However, single-film pre-production revenue prediction is more of a wild west, with bold, uncredible claims lacking independent validation.

Some prediction companies have meaningful offerings, such as nifty tools, relevant advice, and a large number of data points, but many are also in the business of using marketing spin to sell false certainty to unwitting new entrants.

Regardless of the scale of snake oil in this sector, it’s certainly true to say that no one has yet been able to devise a reliable way to spot enough hits, early enough in the lifecycle of a movie, to be able to build a credible business on.

(Side note: If any company in this space wants me to independently verify their claims, then please do get in touch. Seed and Spark did this a decade ago, and I was happy to validate their claims).

How to succeed in business without really predicting

If “We only invest in hits” is not a viable business strategy in film, what is?

Here’s the key idea that allows people to overcome the hurdles we’ve seen so far:

Unpredictable is not the same as random

There is an oft-repeated phrase in the industry from screenwriter William Goldman:

No one knows anything.

The thing is, people have edited the original quote to change its intended meaning. They use it to mean “Everything is unknowable”, whereas the full quote is closer to “Nothing is certain”. Here’s the full line from Adventures in the Screen Trade:

Nobody knows anything...... Not one person in the entire motion picture field knows for a certainty what’s going to work. Every time out it’s a guess and, if you’re lucky, an educated one.

So it’s not true that no one knows anything. In fact, many people know a lot. Hollywood has run the global film industry for over a century with surprisingly little change, given how unpredictable any one movie can be.

This is because while betting on any one particular movie is a risky proposition, there are numerous things people can reliably and successfully do to build a profitable business in the film industry.

It’s about shifting the core question from “How do I pick the winner” to “How do I change my relationship to the risk”.

Below are 18 approaches professionals use to sustainably make money, none of which require predicting a film’s appeal or outcome. Think of these more as aspects of risk management than as a playbook you can just follow by the numbers.

1. Make many films, not one

An interesting quirk of the asymmetrical nature of film returns is that volume converts an unpredictable bet into a dependable distribution. In short, if funding a movie is basically gambling, making a whole bunch of them is basically owning a casino.

The chart below illustrates what happens as you stop betting on one film and start betting on many. The individual films are as unknowable as ever, but suddenly it looks a little more like an actual business.

So if you are fortunate to have access to vast amounts of capital (and nothing less risky or with better upside available to you), here are some things you may wish to consider:

• Build a slate so winners cross-subsidise losers, and you stop caring which is which. This is a core part of the philosophy of Hollywood studios.

• Co-finance slices of many slates, taking 10% of fifty films rather than all of five. A good example was Legendary’s long partnership with Warner Bros.

• Run a low-budget volume factory where one breakout pays for years of misses. Paranormal Activity, made for about $15,000, took $194m, kicking off Blumhouse’s stellar horror-based business model.

• Turn one hit into a franchise, a fresh roll with dice that already landed once. Predicting the release date of a sequel or next instalment of a much-loved franchise is a safer bet than any random film (albeit I don’t think the backers of Joker 2 would have much time for this suggestion!).

Read more

Four geologists at the University of Gothenburg spent more than a decade keeping a list of every film they could find that featured a geologist. Across the 141 movies they tracked, the on-screen geologist died 34% of the time. Usually quite early, and often moments after explaining that the volcano is about to go off.

Erwin asked me:

How does the fate of geologists in movies compare with that of other jobs?

This is exactly my kind of question as it’s core to the work I do creating new movie records for Guinness World Records.

So I turned to the data to see just how slated geologists are. Whether their luck had eroded, their prospects had hit rock bottom, and their odds of survival were, frankly, being taken for granite.

I studied 43,983 movies to track the fate of 328,538 characters who shared 10,112 different job titles and experienced a combined 69,832 on-screen deaths. I can only study the data available in screenplays, film synopses, and plot descriptions, so there’s a certain amount of bias. I have included more about my methodology and data at the end of the article in the ‘Notes’ section, which might be better called ‘Pedant’s Corner’ (or should that be Pedants’ Corner? I mean, it’s not really a corner….)

What are the most dangerous jobs in movies?

Across all the 328,538 characters for whom I could determine both a job title and their fate, 21.3% died. Bummer for them, but great for us, since we now have a baseline. Being a named character in a movie has a 1-in-5 death rate.

It’s three times more dangerous to be a Commando in a movie than an average character. They came out as the ‘Most killed’, with almost two-thirds of them not making it to the final credits. Then, a terrorist, a gunman, a revolutionary and, naturally, a sorcerer.

Half of all movie assassins die, which, for a job that is entirely about killing other people, is a remarkable number of own goals.

Jobs that seem the safest include wedding planner, matchmaker, illustrator, manicurist or sportswriter.

What even is a job?

Some of the entries on that previous chart are not jobs in the traditional sense. For example, one could see serial killing more as a calling than as gainful employment. Other roles, which may not seem like real jobs at first glance, do pass muster, such as those of an informant or a thug.

Also, if there are any young people using today’s data to inform their future career choices (as I imagine many are), then they may feel underserved by discovering that jobs such as commando, terrorist, gunman, and revolutionary have high death rates. They are people whose entire job is violence, and of course, they die, because a film about armed men is usually a film about armed men shooting each other.

So let’s remove the jobs that are centred on violence and those that people do for the love of it. Oh, and wizards and such - I removed jobs that kids today would be foolish to aspire to.

This reveals that the deadliest ordinary on-screen job, once the soldiers and the sorcerers have gone home, is that of a diver, with a 52% mortality rate.

Just in case any of you were thinking of a long career in crime, the data suggests you should aim for middle management. The pay and benefits beat frontline work, but the death rate is lower than that of the criminal C-suite.

Below are the death rates for jobs with organised crime (I left out the disorganised crime roles, as that veers into hobby territory).

It’s less about your job than the world in which you do it

So far, I have been focusing solely on the characters' jobs, whereas we get an even stronger set of survival cues from the movie in which those characters operate.

Unsurprisingly, horror films have the highest death rate, with almost half of all characters (for whom I could determine both their employment and their fate) dying.

It’s not about bad jobs, but bad people

Unlike in real life, villains die way more frequently than good people (38% vs 11%).

On screen, being the bad guy is very nearly a death sentence, which mirrors what the Gothenburg geologists found: 77% of those they called “evil geologists” died. (I am slightly suspicious of just how much harm an evil geologist could do, given the timeframes they are used to working with, but I’m not going to mansplain evil geology to leading geologists).

The traits of a deadly job

It’s at this point that I have to be honest with you - I have very little experience with proper jobs. I have what people kindly call a portfolio career, which mostly means I do several odd things at once. Over the years, my professional work has included:

• Film data analyst

• Running a story agency

• Inventing new world records

• Writing puns for headlines of the horoscopes in The Sun newspaper

• Giving Kanye West tax advice

I like to think of myself as “business feral” - i.e. hard working but barely employable.

So, when it comes to deciding what is and is not a ‘proper job’, I have very little to draw from. Case in point: see below for the job traits I thought we were worth investigating death rates for.

Other things I tested included:

• Money does not save you. A surgeon on $240k a year dies about as often as a cleaner on $34k (correlation +0.02).

• Automation risk does not matter either (+0.08), so the robots and the screenwriters are after different people.

• And the jobs children dream of are mostly safe, the lone doomed dream being that of an astronaut.

The two biggest correlations were:

1. The first is real danger, but only one kind. Cinema has no idea which jobs shorten your life in a non-visual sense (i.e., those with poor working conditions or associated with negative health outcomes). But the jobs that can kill you in a single dramatic accident, the fisherman, the pilot, the miner, the lorry driver, die on screen too (correlation +0.5).

2. The second is gender. The more men do a job in real life, the more it dies on screen (-0.43). Soldiers, pilots and bodyguards die constantly, whereas nurses, hairdressers and secretaries mostly live. It’s such a strong link that job titles ending in ‘man’ are the deadliest suffix in cinema and jobs ending in ‘ess’ are the safest.

So what have we learned?

Nothing useful.

The fact that I’m unemployable also means I’m unfirable. Sorry :/

Notes

I analysed 43,983 feature films and extracted 328,538 named characters. For each one I recorded a stated or strongly implied job, whether they died on screen, and how. I was working from plot summaries and, for a large share of the films, full screenplays.

A character’s job only got recorded when the film bothers to mention it, so the sample leans towards jobs that matter to the plot. This measures the conventions of screen storytelling, not real life. Genre drives the raw body count more than any single job, so the figures are best read as relative standings rather than your personal odds. In my tests, I found that plot summaries missed about 15% of the deaths a full screenplay would reveal, though they get the running order of jobs right.

The geologists were the obvious test, since the Gothenburg team had already counted them by hand. My method puts the geologist death rate at 31%, against their 34%, which is close enough to trust the wider net.

Eleven jobs turned up in twenty films or more without a single death between them, which seemed too good to be true, so I had each one stress-tested for counter-examples. About half cracked (a circus impresario mauled by a bear, a studio boss with his throat cut, a ski instructor lost to an avalanche, a chambermaid strangled in the billiard room, and an ice hockey player). But five held firm, with no verified on-screen deaths at all: wedding planner, matchmaker, illustrator, manicurist, and sportswriter.

The original study by Sturkell, E., Sjöqvist, A., Björklund, L. and Johnsson, A. (2026), ‘Geologists on the silver screen, the sequel’ was published in Geology Today, 42(2), pages 82 to 86, which is open access and worth a read. It is the source of the 34% geologist death rate and the 77% ‘evil geologist’ figure, and of three lovely examples: Dr No (1962), where 007 shoots the evil geologist, Prometheus (2012), where one is killed by aliens, and Dante’s Peak (1997), which somehow finds room for seven heroic geologists at once.

‘Dangerous in real life’ comes in two flavours, and they behave very differently.

• For deaths actually on the job I used the US Bureau of Labor Statistics Census of Fatal Occupational Injuries

• For the jobs that non-visually shorten life overall I used UK Office for National Statistics occupational mortality.

• Pay is from the BLS Occupational Employment and Wage Statistics

• The share of each job done by women from the BLS Current Population Survey

• Automation risk from Frey and Osborne’s ‘The Future of Employment’ (Oxford Martin School)

• Public trust, which predicted nothing, is from the Ipsos Veracity Index.

The ‘dream job’ measure was my own best calibration to childhood aspiration rather than hard official numbers. All correlations are rank-based (Spearman), so they describe whether the order matches, not the exact sizes.

That fight, though, was almost entirely about television, not film. The most pernicious version of packaging lived in TV series, where an agency could earn as much as, or even more than, the very people who created the show, and where the agency had little reason to push for higher writer pay, because its money came from the package fee rather than the writer’s salary.

Packaging fees may have ended (at least for television writers), but today I want to look into two questions:

1. How prevalent was/is packaging among movies?

2. Did it really end?

What is packaging, and how do we track it?

Despite being carried out by a small number of well-known agencies that represent the world's most famous people in the biggest movies ever, packaging is relatively hard to study.

For today’s research, I have built a fairly good proxy, but before I reveal the numbers, I need to explain a little more about what packaging is and how we might detect it.

Formal packaging is a specific deal talent agencies make with Hollywood studios. Rather than take its usual 10% commission from each client, an agency assembles the talent and charges a fee paid by the studio. In television that settled into the “3-3-10” structure, which broke down as:

• 3% of each episode’s licence fee, paid up front as episodes are made (roughly $15k–$75k an episode).

• 3% of the licence fee, but deferred until the show hits “net profits”. Due to Hollywood accounting, this would often not be triggered.

• 10% of the back end (syndication and foreign sales) over the life of the show. This is where the real money for the agencies was, as even a small number of mega hits could pay a lot for decades to come.

Here is a helpful Hollywood Reporter article which explains more.

The whole notion of packing fees skewed the incentives for agencies, and many writers felt it meant they were working against the best interests of the very clients they were meant to represent.

In his usual measured, calm style, David Simon of The Wire wrote:

Packaging is a lie. It is theft. It is fraud. In the hands of the right U.S. Attorney, it might even be prima facie evidence of decades of racketeering.

A slightly more measured came for the estimable John August:

When your agent is also the buyer, that's the clearest conflict of interest imaginable.

In response, the agencies didn’t have much more of a defence than “We like the money packaging gives us, and we want to keep doing it”. Jeremy Zimmer, CEO of UTA, said in a letter to their writer clients:

Over the past couple of decades the complexity of the business and the opportunities that have been presented have allowed agencies to decide on a path and pursue it…. We aren’t saints. We are tough businessmen and businesswomen who happen to love artists.

Karen Stuart, executive director, Association of Talent Agents told Deadline:

This development is ironic given that the Guild itself has agreed to the legitimacy of packaging for more than 43 years.

In April 2019, the Writers Guild told its members to fire their agents in protest, and over 7,000 did. After a bit of bluster, the agencies caved and signed new franchise agreements one by one. Packaging fees on Guild-covered projects were switched off from 1 July 2022.

Searching for packaging-like behaviour in film

Packaging fees were overwhelmingly a television business, built on the long tail of syndication money that almost all films lack. The Writers Guild calculated that 87% of scripted series in 2016-17 were packaged, but there has never been an equivalent figure for film.

Film packaging also works differently. It tends to live in the independent film world rather than the big studio system, and it almost always starts with a script. A writer takes that script to their agency, which then tries to attach a director and a star, and the resulting package goes out to raise financing at markets like Cannes and the American Film Market.

And it is not only agencies that put these bundles together; independent producers and production companies assemble the same script-director-star packages too.

I don’t have inside knowledge of all the deals conducted between agents and studios, neither the official ones nor the murky world of favours and influence trading. But I can look for patterns in the data.

I built a database of every movie released between 2010 and 2025 with at least 50,000 IMDb votes. I then narrowed in on the film’s core creatives, which I have defined here as the director(s), writer(s), and the top three credited cast members.

I excluded producers, not because I don’t think they’re creative, but because (a) there are a large number of empty producing credits (see more on that here), and (b) producers are more likely to represent themselves.

Finally, I sought to determine which talent agency represented each creative 18 months before the film's release. If three or more of those key creators shared a single agency, I flagged it as a “One-agency Film”.

Sharing an agency doesn’t automatically mean the film was officially packaged. This research has a lot of fiddy details and caveats, which are in the Notes section at the end, along with my methodology.

Have one-agency films increased?

One-agency films have grown dramatically, from around 10% of mainstream films in 2010 to around a quarter in the current decade.

The single highest year is 2023, when nearly three in ten big films had three of their key creators at one agency.

A rising line on its own does not prove agency action on its own. Last year, I looked at which Hollywood agencies dominate the representation of film directors and actors and found that just over half of the top actors and a third of the top directors are with one of just four agencies.

If the big agencies simply represent more people than they used to, you would expect more of the one-agency film pattern we’ve just seen to emerge through chance alone.

Could this be a coincidence?

For each year I took the actual market share of each agency among that year’s creators, then calculated how often a random four-person team would land three or more people at the same firm purely by chance.

Put simply, if everyone picked their agent out of a hat, weighted by how big each agency really was that year, how many one-agency films would you get?

The chance line sits between 7% and 11% for the whole fifteen years. The agencies did grow, and that lifts the baseline a little, but only a little.

This leads us to conclude that almost all of the rise in the observed rate is due to deliberate actions by some people.

In the early 2010s, one-agency films had about a 1.4-times the chance level. Today, they run about 2.6 times higher.

So the concentration is not only larger, it is harder to explain away as a coincidence than it used to be.

Which agency?

Almost all of the clustering comes from three firms: CAA, WME and UTA. The rest (Gersh, APA, Verve, Paradigm) turn up on a film or two a year at most.

Two things stand out to me:

The first is CAA’s 2023 spike, at nineteen one-agency films, the highest for any firm in any year of the window. That is almost certainly the CAA-ICM merger landing: CAA absorbed ICM Partners in June 2022, and a slice of what would have been ICM clusters a year earlier became CAA clusters once the books combined. By 2024-25 CAA settles back to seven or eight a year.

The second is that WME wins more years than that single CAA spike suggests. It tops 2017 and 2024, among others. Across the full fifteen years, the two are much closer than 2023 alone implies, with CAA on 138 cluster-films and WME on 117. UTA holds a steady third, strongest in animation and family-franchise work.

All three have changed hands during this period. CAA went to François-Henri Pinault’s Artémis, and WME to Silver Lake, which rebranded it the WME Group.

The writers, not the stars

When I split the overlap rate by role (how often a director, a writer or a top-billed actor shares an agency with at least one other key creator on the same film), writers cluster most, rising from 47% in 2010 to 61% in 2025.

If you want to find a one-agency film, look at who wrote it rather than who is on the poster.

This is exactly what we would expect if feature packages really do begin with a script. If the writer is the first element in the room, they become the anchor around which an agency builds the rest of the film.

The Writers’ Guild fight shows up here too, faintly. The writer line dips and stalls around 2019-21, exactly when the action was live, and a chunk of the writing population had no agent at all. Then it climbs again, and by 2023 it is at a fresh high.

This is what happens when you target just one aspect of a problem. It’s like when you say to a kid, “Don’t let me catch you doing that!” and they reply, “Ok, I won’t let you catch me”.

In this instance, the Writers’ Guild said, “You shouldn’t be charging a fee for not working against the best interests of your clients” and the agencies replied, “Ok, we’ll stop invoicing”.

Notes

Today’s research examines 2,058 films released between 1 January 2010 and 31 December 2025 that received at least 50,000 IMDb votes. Of those, 1,989 had at least one key creator whose deal-time agent could be resolved. The films that drop out are mostly international productions whose creators have no US agency footprint, or films whose entire teams were first-timers not yet on the signing record.

The tracked ‘key creators’ were the credited director, all credited writers, and the top three billed cast.

I also dropped anyone whose recorded date of death predates the film’s release, which removes source-material credits such as Bram Stoker for Nosferatu or Mary Shelley for Frankenstein.

I use the release date minus eighteen months as the proxy for when a film was assembled. My per-person agent timeline behind it contains 47,562 records of (person, agency, date range) across 18,757 people, built from trade-press signing announcements in Variety, The Hollywood Reporter, and Deadline dating back to 2010, then overlaid with a May 2026 representation snapshot.

In today’s research, the term ‘agency’ refers only to talent agents. People often have separate representatives for commercials, voice, theatre and literary work, plus separate managers, lawyers and publicists. Counting all of those would multiply the clusters without telling us much.

The chance baseline is calculated using a binomial distribution. For each agency, take its share of all deal-time-repped creators in the cohort that year, work out the probability that a four-person team would land three or more people at that firm by independent choice, then sum across firms. It is recomputed every year, so the growth of the big agencies is already accounted for.

It’s worth repeating that sharing an agency is not automatically the same as being packaged. This measures single-agency concentration at deal-time and the odds that it is a coincidence. It does not, and cannot, confirm a packaging deal or fee on any individual film. Historical and corporate context comes from the trade-press record and the Writers Guild’s own agency agreements.

’ll be uploading new video essays every couple of weeks throughout 2026, so make sure you’re subscribed on YouTube to get notified when another drops (also, subscribing will really help me out with the Algorithm Gods, so it would mean a lot to me - thank you!).

https://www.youtube.com/@TheFilmDataScientist

And to help explain the channel, I also created a short channel trailer:

Rather than celebrate this achievement, some have decided that it’s just not possible and have built a fringe conspiracy theory that he did not actually direct the film. What a world we live in, eh?

Kane himself seems amused by the patently false rumour, and commented on it exactly how you might expect a 20-year-old to - via a pinned YouTube comment in which he said:

This is actually all true. They don’t tell you this, but 96% of all movies released in North America and Europe are actually directed by the same person. They call him ‘The Older Gentleman’. He has had all of Hollywood in a chokehold for decades now.

I looked at the directors’ ages behind 4,690 credits to see how old they were when their movie was released.

Rather than looking at all movies made, I focused on those with at least 50k votes on IMDb. This enables meaningful coverage (many early-stage directors have biographical data available) and lets us focus on professional, working directors.

For those wanting a sense of how famous a film has to be to be included, this filter just about includes the 1998 Very Bad Things starring Cameron Diaz, the 2007 horror film Teeth, and the 1995 Jackie Chan beat’em up Rumble in the Bronx.

Forties for the win

Across the dataset, the median director age at the time of release is 45. Half of all credits fall between 39 and 52. The distribution is right-skewed, which is a data nerd way of saying a few very old men are pulling up the average.

While this is interesting already, what's under the surface of these numbers is even more revealing.

Have directors always been this age?

You might have expected the headline average to have slowly drifted upwards, given that the UK and US working populations have aged over the past forty years, and many creative industries have aged with them. But that’s not what we see in the data.

The second-highest decade for directors’ ages was the 1950s, when the median age was 47. This is the tail end of the studio era, when the people running widely-seen films were largely those who had been running them since before the war. Alfred Hitchcock, William Wyler, John Ford and George Cukor were all in their 50s and 60s, and Cecil B. DeMille was directing into his 70s.

Between the 1950s and the 1980s, the median fell five years, from 47 to 42. This is the studio system’s collapse, the French New Wave’s influence on the Americans who noticed it, the rise of film school as a career route, and a New Hollywood generation who broke in early. Peter Bogdanovich, Martin Scorsese, Francis Ford Coppola, Steven Spielberg, Brian De Palma and George Lucas were all directing widely-seen features in their 20s and 30s. Directors got younger because the gates opened.

This century, we’ve seen a reversal of the trend, and by the 2020s, the median director of a widely-watched film is seven years older than their 1990s equivalent.

So it’s true to say that film directors are getting older… but only if you compare now to thirty years ago. And we’ve now edged just past the old 1950s high, the oldest the director’s chair has ever been

The changing faces of film directors

As new types of people were being given command of commercial films, we have seen the director’s age change. But while one aspect evolved (i.e. ‘hippy types’ took over from the studio old hands), another did not - they were still mostly men.

The 1990s saw an increase in female directors, and in more recent years, we’ve seen another increase in diversity (albeit far from parity). You might reasonably ask how this change affected the average age of directors.

Male directors have got much older (from a 1980s mean of 43 to 50 today), but the average age of female directors has barely moved, hovering around 44 for the past two decades

This means we need to update the finding we stated earlier in the article. Rather than concluding simply that “Film directors are getting older”, we need to say instead that “male film directors are getting older”.

Women are pulling the overall trend down, not up. Without the male credits, the 2020s mean would sit at 50 rather than 49.

Are old people hoarding all the cash?

As with wider society, it is indeed true that most of the money is given to, and held by, the most age-challenged folk.

Joking aside, this actually makes sense. Bigger budgets are generally given to people who have proved their capabilities on previous projects, so they are more likely to be older. In addition, the types of films that can raise the biggest budgets are also those that tend to pick their key creatives from established networks.

Which genres are run by younger directors?

As well as Backrooms, last month also saw the release of box-office smash horror movie Obsession, directed by 26-year-old Curry Barker. This is a slightly more common sight, as horror has the youngest average director age of all genres, at just 40 years old.

Some of this is a function of lower budgets, and it mirrors the audience demographic, in which horror has the youngest adult audience, whereas biographies and dramas have among the oldest.

Further reading

Here are other projects I’ve conducted on age demographics:

• Why are actors getting older?

• Is the over-50 action star boom peaking already?

• How old are horror movie audiences?

• Do older audiences want to watch older actors?

• How do movie genre tastes change with age?

• What films are older cinemagoers watching?

• Can the film industry halt the decline of young audiences in cinemas?

• What percentage of movies are written by women over 40?

Notes

Today’s research examined every feature film with at least 50,000 IMDb votes that was released theatrically since 1930. That gave me 4,690 director credits across 4,203 films, of which I could find reliable birthday years for 95%.

I worked out the age by subtracting the year of birth from the film’s release year. This is coarse (i.e. a director born in December releasing in January will be out by one year) but at decade-scale trends this does not matter.

Production budgets are from OpusData (The Numbers) and are nominal, not inflation-adjusted.

For the avoidance of doubt, this is directors and co-directors, not the second-unit, assistant, art, and casting directors.

For most of film history, people publicly identified with one of two genders, so I have mirrored this binary approach in reporting the numbers. The number of people this excludes is a rounding error, so not relevant to today’s topic.

Do you know the percentage of a film festival’s budget that comes from entry fees? I have heard that it was 9% - what have you heard?

Thirteen years ago, when I interviewed 523 film festival directors, the average was under 14.8% of income.

However, when the Raindance Film Festival opened its books to me three years later, its submission-fee income accounted for 35.6% of its total income.

So I didn’t have a clean figure to provide to Barton off the top of my head. I went digging around for data sources. In the US, most film festivals are non-profits, and non-profits file a public tax return, the Form 990. Buried inside it, on the Statement of Revenue, the festivals list where their money came from.

For example, in 2023, the Portland Film Festival took in $307,194 in income, 82% of which came from fees.

So, we can already see that the answer will differ widely between festivals. Let’s take a systematic approach, looking at the details of two sources of data:

1. 2013 Festival Director Interviews. The answers given by film festival directors in 2013. Each provided a single, self-reported figure for their festival.

2. 2026 Tax Records Analysis. The tax records of US-based non-profit film festivals. Most organisations lump all income into a vague bucket marked “programme revenue”, but a minority break it out into a separately labelled line item, “Submission fees” or “Film entry fees”. Out of around 2,500 candidate organisations, 57 itemised their fee income cleanly enough to use, giving 266 festival-years of data.

The short answer is “it varies enormously”

Using the new data, I found that three-quarters of film festivals earned less than 11% of their income from fees.

The smallest share from fees was just 0.2%, and we’ve already seen the largest, at 82%, for Portland.

The average percentage across these festivals and all the tax records was 10.7%. The median was 4.4%.

One thing worth noting is that the smallest festivals are the ones most likely to rely on fees, and yet they are also the least likely to have detailed tax records.

Has it changed much?

Using the tax records we can look back over the past decade. This reveals that the median festival has fluctuated between 3% and 7% over this period.

How are film festivals funding themselves?

At this point, you might be wondering how festivals stay afloat, if filmmaker fees are such a thin slice.

Festivals earn their money from three types of income:

1. Submission fees paid by filmmakers

2. Donations, grants and sponsorship

3. Commercial income, mostly ticket sales, but some do have merch.

Fees are the smallest of those three types. Below, I’ve broken it down by festival age and put both datasets side by side: the 2013 director survey and the 2026 tax records.

The oldest festivals rely on fees the least. Veterans take just 7% of their income from fees in both datasets because they’ve had the time to build donor and regular audiences that younger festivals are still chasing.

My reading of the data reveals four kinds of film festivals:

1. The donor-funded are the mainstream, getting the bulk of their income from grants and gifts and only a few percent from fees.

2. The sponsor-and-ticket festivals are the biggest, covering their costs with a mix of donations, box office and sponsorship.

3. The small-and-fee-leaning are tiny operations scraping a third from donors and leaning on fees for close to a fifth of their income.

4. The fee-funded are tiny, donor-less, and running on filmmaker money for three-quarters or more of their income.

The biggest of the “Donor-funded” cohort is Sundance. Its audited statements reveal that fees account for only around 2% of total income. Sundance processes roughly 15,000 submissions a year and earns far more from sponsors, donors, and ticket sales, so the $1 million it collects from submission fees is a rounding error on its books.

Cork International Film Festival, Ireland’s oldest, is a useful contrast as a public-arts festival rather than a US non-profit. Its published accounts have held steady between 8% and 11% over six years.

How much money are we talking about?

I have focused on the percentage of fees earned, rather than the dollar amount, for two reasons. Firstly, that was the focus of Barton’s initial question, but also because I think it’s a more meaningful number.

However, I’m sure I will be asked about the numbers at play, so here’s a quick overview.

Last year, I studied the cost of a single entry to a film festival, discovering that the average fee ranged from $30 (for an early-submitted short film) to $87 (for a late-submitted feature film).

In today’s new research on tax returns, the festival that reported the highest income from fees was Santa Barbara, at close to half a million dollars. But Santa Barbara is a multi-million-dollar operation, so that half-million is only about 4% of its income. (There will be many festivals earning more, but their tax reporting prevents us from knowing the figures).

The Dumbo Film Festival has a much lower total but is at the other end of the spectrum. They reported that in one year all of its $51,964 funding came from fees, and 72.7% was paid out to just three people.

Below are the festivals that had the largest declared income from fees.

So what of ‘scam’ festivals?

Filmmakers have been warning each other about ‘scam festivals’ for well over a decade. These are events that exist mainly to trick filmmakers into paying to submit, under the false premise that the festival is well-respected, useful, or even real.

The concern is neither new nor fringe. Back in 2019, Screen Daily reported calls for a festival regulator after a run of scam events, and noted that FilmFreeway had already delisted 42 of them. The Hollywood Reporter has documented how below-the-radar festivals prey on struggling filmmakers, including a horrifying story of a writer-producer who borrowed money to fly from New York to London for her ‘UK premiere’ and found a near-empty room. One Rhode Island showcase ended up branded the ‘Fyre Festival’ of film festivals.

It helps to picture a spectrum. At one end sit the outright frauds, which take your fee for an event that never happens. At the other side are the ‘award mills’, which are technically real but exist to manufacture laurels. They run rolling monthly deadlines, nominate almost everyone, spin up a dozen city-branded editions of the same event, and then sell you a certificate, a trophy or a ‘social media package’ once you’ve ‘won’. The laurel, not the screening, is the product.

Sadly, the tax data I’ve been using can’t really tell the difference between a ‘real festival’ and one that just farms fees. And to be fair to the small festivals at the top of my charts, a festival funded almost entirely by fees isn’t necessarily doing anything wrong. It may simply be a small event in a town with no arts money and no wealthy donors, funding itself the only way it can.

Similarly, the fact that most of the income goes to a small number of people is not automatically suspect. We have all met the person who makes a community event happen through sheer dedication and hard graft. If they secure venues, volunteers and resources for free, then paying them for their time is good value, not a red flag.

So how do you tell the two apart? The community has settled on a familiar set of warning signs: dozens of award categories, near-automatic nominations, vague or shifting venues, screenings that turn out to be ‘online only’, and a flurry of upsells the moment you’re selected.

There are good rundowns of the red flags from Noam Kroll, Submitting to Film Festivals and Show Me Shorts. The single most useful habit is to do your due diligence before you submit your film each time. Check that the venue exists, look for photos and press from past editions, and ask what will be screened if you’re accepted.

The platforms have, belatedly, started to respond. FilmFreeway now runs ID checks, demands venue proof, and holds a new festival’s first-year payouts until it can prove the event actually took place. Filmmakers have petitioned for more, and bodies such as the Universal Film Festival Organisation have launched specifically to stamp the practice out.

The voice of the festival directors

I don’t want to end on a negative note, as the vast, vast majority of film festivals are working hard and doing well.

A filmmaker who is starting to submit to festivals might be forgiven for initially decrying the amount of money it costs to submit to festivals, but once you dig into what it takes to put festivals on, the fees seem far more defensible.

Running a film festival is a lot like making a film. I.e., it makes no rational sense, so the only people who do it are those who are most passionate, most resourceful, and who really believe in the mission.

In preparing today’s article, I reached out to a number of the festivals in my dataset, especially those leaning most heavily on fees, to ask how it looks from their side of the desk.

Josh Leake, Executive Director of the Portland Film Festival, rightly pointed out that tax filings only capture cash, not the volunteer and community support that keeps a small festival running. He said:

The submission-fee revenue can appear larger on the filing than it would if one were looking at the festival’s broader support base as a whole.

Jeff Ross of SF IndieFest explained why a festival’s fee reliance can climb without anyone raising a single fee.

Before the pandemic, our revenue was mostly from ticket sales, with some grant money and submission fees making up a much smaller percentage. The percentage of revenue from submission fees has gone up, but that’s mainly because our revenue from ticket sales has declined so much. The overall event budget has declined about 50% since 2020. We don’t have sponsors, donors or a membership program. We are truly a grass roots DIY arts org.

At the other end, Leslie Raymond of the Ann Arbor Film Festival, whose fee reliance has actually fallen over the decade, said:

Submission fees are an important part of our funding mix, but our goal has always been to build a sustainable festival through a diverse range of revenue sources rather than relying too heavily on any single one.

And Kitty Boone of Aspen Film showed how one organisation can do both. Its feature festival is invitation-only, with no submission fees at all, whereas its shorts festival, Aspen Shortsfest, charges one, because:

…we receive well over 3,300 submissions, [and] it requires a lot to program a festival such as this.

The same themes came through thirteen years ago, in the anonymous answers to my original survey:

“It is a labor of love. No one is getting rich off your entry fees.”

“Generally, I think filmmakers believe that festivals have more money than they have… there usually is no ‘profit’ to share. The vast majority of film festivals barely get by.”

“There is very little funding, and small festivals like us rely on the nominal entry fees to let us run a festival.”

“We know people don’t like paying submission fees, but without them our festival (and I assume many others) would not be able to make ends meet.”

Notes

I built the festival list via ProPublica’s Nonprofit Explorer and the IRS Business Master File. That gave around 2,500 candidate organisations. I matched them to the IRS e-file index and read the underlying Form 990 XML directly. 57 festivals itemised submission or entry fees on their own line (Form 990, Part VIII, line 2), giving 266 festival-years across the 2015 to 2024 tax years. The denominator throughout is total revenue (Part VIII, line 12).

I only used a festival year where entry fees sat on their own labelled line. Where a festival bundled them into something mixed, like “Tickets and entry fees” or a generic “Festival fees”, I left it out rather than guess.

These are fees as booked (i.e. gross, not net), before FilmFreeway’s commission of around 12%, payment processing, sales tax or fee waivers. The cash retained is somewhat lower, so if anything, these figures slightly overstate reliance.

Organisations below the filing threshold submit a Form 990-EZ or a one-page postcard, neither of which itemises fees. Of 609 festivals that filed the short form, only about one in eighteen mentioned submission fees anywhere, and never as an isolable number. They are exactly the festivals most likely to live on fees, so the true long tail sits higher than this sample, not lower.

Sundance and Cork come from their own published accounts, not the IRS data. Sundance’s audited statements carry a “Submission” line and Cork’s accounts, filed as an Irish charity, carry a “Submission Fees” line.

A few readers reached out to ask whether I had taken the films’ quality into account. I had, but what stuck with me was how many people clearly assumed that a famous cast meant a better film. The cynic in me assumed the opposite!

Let’s briefly consider the two perspectives, and then see which the data reveals to be correct.

Fame = quality?

You can see why someone might think that famous actors would be in better films.

Famous actors are, on the whole, richer and far better supported than those still on the way up. They have readers, managers and agents whose job is to find good scripts. Not only can they seek them out, but they will also be constantly offered more of them because most filmmakers want/need a famous name in their film.

They also have power - Tom Cruise can pretty much star in almost any film he wants (if he is willing to be flexible on his fee, of course).

So it might be reasonable to conclude that more famous actors tend to make better films.

Fame ≠ quality?

But there is a counterargument that famous actors are not trying to make good films. They are trying to make films that are good for them. This can be a very different thing, tied to their brand and money.

When asked about his role in panned Jaws: The Revenge, Sir Michael Caine famously said:

I have never seen the film, but by all accounts, it was terrible. However, I have seen the house that it built, and it is terrific.

It’s a reasonable assumption that the more famous the cast, the bigger a film’s budget will be, but I doubt many people instinctively believe that bigger budgets automatically mean better scripts.

Then we have to consider that the biggest stars tend to work on films with the most competing forces around them, such as marketing, release-date strategy, and merchandising.

None of these forces aims to improve the film, and some directly pull against it.

Scores ≠ quality?

One thing to note upfront is that we cannot actually measure whether a film is good. We can only measure whether people liked it. Today I’m going to be using two quick proxies:

1. Metacritic score, which averages professional critics’ reviews into a number out of 100.

2. IMDb user rating, which captures the general audience’s verdict out of 10.

So in truth, we’re tracking reception rather than art, but they are the best signals we have of how a film landed. At the very least, they let us see which factors correlate and which do not.

I crunched the numbers on 2,756 feature films released over the past quarter-century to see if there is a connection between how audiences and critics rate them and the cast’s star power. The details of my methodology are in the Notes section at the end.

A famous cast is a weak sign of quality

The core finding is that fame is a pretty bad proxy for whether a film is likely to be any good.

Films led by a top-tier name score ever-so-slightly higher than those led by a non-famous actor, with both audiences and critics, but the gap is small and only with the most famous names.

Top-tier-led films average 6.65 out of ten with audiences, compared to 6.28 for ordinary-led films, and a similarly modest step with critics. Once you account for the film’s budget, genre and era, the famous cast still helps, but only barely.

It depends heavily on the genre

That faint average hides a sharp split by genre. A famous cast is a meaningful hint of quality in some kinds of film and no hint at all in others.

In adventure, a more famous cast tends to go with better-reviewed films. Everywhere else, the link is weak, and in drama, romance, and horror it is essentially zero, even tipping slightly negative among critics. These are exactly the genres that live on concept, script, and word of mouth, where a familiar face adds little and can even raise suspicions that a film is relying on its casting.

Over the entire period, the cast-quality link has remained weakly positive, neither strengthening nor collapsing, hovering between +0.08 and +0.22 across eras.

Which actors are a good sign of quality?

The link is much stronger with some actors than with others. Some actors appear, again and again, in films that audiences and critics both rate highly. Others have built long careers in films that neither does.

The scatter plot below places each of the 2,730 famous actors I looked at on a chart of average IMDb and Metascore for their past films. They fall into four camps:

• Everyone’s favourites are in the top right, including Leonardo DiCaprio, Tom Hanks, Cate Blanchett, Denzel Washington, Christian Bale, and, among newer arrivals, Florence Pugh.

• Critical darlings are in the top left, and include Greta Gerwig, Robert Pattinson and Kristen Stewart, whose choices lean towards the kind of film reviewers admire more than the multiplex crowd.

• Crowd-pleasers sit in the bottom right, containing actors whose films are appreciated by audiences but much less so with critics, such as Will Smith, Ryan Reynolds, Sandra Bullock, Liam Neeson and Gerard Butler.

• The… er… others. In the bottom corner, we have actors whose films are poorly rated by both audiences and critics, such as Adam Sandler, Nicolas Cage, Bruce Willis, Dwayne Johnson, and Jason Statham.

Adam Sandler appears completely unbothered by how his movies are reviewed, saying in 2013:

I didn’t get into movies to please the critics. I got into it to make people laugh and have fun with my friends.

Others in that final quadrant draw a clearer distinction between the film and their contribution to it. When IndieWire asked Nicolas Cage about the run of films he made to pay off a multi-million-dollar debt, he said:

Even if the movie ultimately is crummy, they know I’m not phoning it in, that I care every time.

If you do want to seek out good films based only on the famous names starring in them, below are the people with the famous names who have the highest correlation with quality:

1. Leonardo DiCaprio

2. Andy Serkis

3. Florence Pugh

4. Philip Seymour Hoffman

5. Warwick Davis

6. Carey Mulligan

7. Viggo Mortensen

8. Clint Eastwood

9. Tilda Swinton

10. Helen McCrory

11. Benedict Cumberbatch

12. Tom Hanks

13. Andrew Garfield

14. Quentin Tarantino

15. Christopher Lee

16. Domhnall Gleeson

17. Brad Pitt

18. Tom Hiddleston

19. Léa Seydoux

20. Tom Cruise

Bonus Finding

As I had the data to hand, I took a quick look at what factors do seem to matter.

The strongest single signal is the film’s runtime, albeit we’re still not in a ‘usefully predictive’ space. R-rated, adult-aimed films score higher than gentler ratings, and films where the director also wrote the script score higher, especially with critics.

Everything I looked at sits in the shaded zone, where the effect is small. I have sorted them by signal strength, meaning that the cast’s fame is among the least predictive.

The fact that longer films score higher is pretty consistent, rather than just the average being pulled up by the canons of Christopher Nolan and Martin Scorsese.

Notes

Today’s research looks at 2,756 feature films released between 2002-24 with:

• A budget on record,

• A worldwide box office figure

• A lead whose fame could be measured before release

• Both an IMDb user rating and a Metacritic score.

The actor chart uses a broader set of 2,730 prominent actors, each with at least fifteen rated theatrical films released since 2000.

Cast fame is a relative score for each film’s lead, built from publicly visible popularity signals gathered across several websites in the year before release, ranked against films of the same era.

Quality is the IMDb audience score (out of 10) and the Metacritic critics’ score (out of 100). Both are gathered after release, so they describe how a film was received rather than some inherent quality fixed in advance.

To be featured on the scatter plot, the actors had to have been among the 1,000 most famous actors at some point and to have appeared in at least 15 films released since 2000 that had both an IMDb audience score and a Metascore.

The signals chart shows a regression of each score on cast fame, budget, runtime, producer count, franchise status, adaptation, rating, whether the director also wrote it, animation, genre and era. Bars show each factor’s strength on a standardised scale, allowing them to be compared directly. With several thousand films, even small bars are statistically real rather than chance, so the shaded zone simply marks effects small enough to matter little in practice. Runtime and budget are associations, not proven causes.

The actor chart leans towards prominent, mostly English-language leads with enough films to be measured. Films that neither audiences nor critics have scored are absent, which skews the set towards titles that got a real release.

The list of the highest correlated stars at the end focuses on actors (not pure voice actors) and the more famous names.

Further reading

I’ve written on the topic of quality a few times over the years, including:

• Whether film length signals quality,

• Which actors appear most in excellent and awful films,

• Whether fewer producers make better films,

• Whether writer-directors make better films,

• Whether the timing of reviews predicts quality.

It might be the oldest argument in the film business, and the two sides go like this:

• One camp says a famous face is the closest thing to insurance a producer can buy. Audiences seem to see them as consistent brands in which every film is a prototype and genre labels are fuzzy at best.

• The other camp points out that the famous names are only there to protect the risk-averse gatekeepers within the production and distribution supply chain. Once movies hit audiences, they just want it to deliver and be entertaining.

While Obsession may have provided a powerful visual, the plural of anecdotes is not data. I went back to the data, studying the box office performance of 3,052 films this century, in relation to the fame of their lead cast (more details in the Notes section at the end of the article).

Does having a star help your film succeed?

Given that we can’t know how much an actor is paid upfront (or how much of the income they also get), we can’t talk about true value for money. For every star, there will be a price at which their involvement is a net positive to the bottom line, and after which they cost more than they help.

But we can measure the difference in performance between films with famous, semi-famous and non-famous names. For this, I am using the worldwide box-office multiplier relative to the budget. Let’s not worry too much about the exact numbers; instead, focus on the difference between the cohorts.

Films led by an actor in the top 10% of fame for their era grossed a median of 2.54 times their budget worldwide. Films led by someone in the 50th to 75th percentile, the group I am calling ‘Low Fame’, grossed 1.91 times.

So we can see that fame does have an effect.

The difference between ‘not famous’ and ‘kinda famous, I guess’ is nothing. It’s only as we start to talk about actors in the top quarter that we see an uplift, with the largest impact on the biggest name.

Given that those are also the highest-paid actors, it’s fair to assume that some of that uplift doesn’t go to the producers, studio, or investors, instead ending up in the pockets of the actors (and their agents, managers, stylists, doctors, and spiritual gurus).

Are there other ways having a famous name helps?

So many of the decisions in the film industry are made in response to perceived risk. Some can be seen as professionals making smart, informed decisions, but in a large number of cases, it’s people acting instinctively, reactions rooted in fear. So it makes sense that having a famous name can feel reassuring.

So while the previous graph SHOULD be the most important one, I suspect the one below is more interesting to film professionals. It shows that having a famous name helps reduce the risk of an embarrassing box-office failure.

A star does not guarantee a hit, but the floor under the downside is noticeably higher.

Where fame pays, and where it barely matters

But these effects are not spread evenly across genres. Obsession appeals to a younger, horror-friendly crowd.

When we split our data by genre, we find that while an A-lister is associated with higher returns across every major genre, the size of the effect varies widely.

The biggest payoff is in romance, where top-tier-led films return a multiplier of 3.82 versus 2.08 for lower-tier leads. A romantic film is largely a question of whether audiences want to spend two hours in a particular person’s company, and a recognisable, well-liked lead is close to the whole product.

Comedy, drama, thriller and adventure sit in a tighter band of roughly half a multiplier.

The genre where fame does least is horror, where the gap is only about 0.25x. This is the same pattern that Bruce Nash and I found when we looked at 877 independent films and the odds of a theatrical release a decade ago. Horror was the genre least dependent on a name and most dependent on concept.

The franchise does most of the lifting

Eagle-eyed readers may have noticed I didn’t comment on the effect of action in the section above. Action had the second largest effect of a star, after romance. However, it is worth asking which famous leads those are actually appearing in. Increasingly, the answer is a franchise.

The share of top-tier-led action films that belong to an established franchise has climbed from around 40% early in the period to about 79% in recent years.

In top-tier action, films in a franchise return a median ratio of 3.47, while those that are not return 1.99. The famous winners are nearly all franchise instalments, from Spider-Man: No Way Home and Jurassic World to The Hunger Games and Aquaman. The famous flops are the films that asked a star to carry a brand-new idea on their name alone: Jonah Hex (less than a quarter of its budget back), Argylle and Mortdecai. The famous name and the franchise usually travel together now, and when you can separate them, the franchise is doing most of the lifting.

A study from 15 years ago concluded that the revenue gap between star and non-star films was driven almost entirely by those films having bigger budgets and wider releases, not by audiences turning up for the face. It is also why some people contend that intellectual property is killing the movie star, and why studios and actors openly disagree about who is actually generating the money, the person or the logo.

Two bonus findings

Before we go, I want to share two minor findings I came across along the way that I found interesting.

The first is that the value of a famous name is stackable:

• Films with no top-quartile fame grossed a median of 2.04 times their budget.

• Add one famous lead, and that rises to 2.30.

• Put two genuinely famous names at the top of the call sheet, and it reaches 2.91.

I hear a few people in Cannes this year complained that while they already had a famous name in their proposed film, sales agents and distributors were not interested until they added a second famous name. While I appreciate how frustrating that must be for the filmmakers, the data seem to support what these sales professionals have been saying.

Secondly, it’s not just how famous the actors are in the moment; it’s also about the wider story of their fame.

Among leads who were already famous, I split them by whether their public attention was rising, steady or falling in the year before release.

Rising stars outperform. At the same level of established fame, films whose lead was climbing returned a median 2.69, against 2.28 for steady names and 2.21 for those whose moment was fading.

Catching an actor on the way up, when audiences are actively getting more interested in them, beats booking an equally famous name on the way down.

Notes

Today’s research examines 3,052 films released between 2000 and 2025, each with a recorded production budget, a positive worldwide box office, and a lead actor whose fame could be measured before release.

Budgets and box office come from The Numbers. The sample leans towards English-language theatrical releases, because those are the films whose budgets are publicly reported. There is a growing (and worrying) trend of reduced budget availability, which I’ve written about before.

In addition, today’s research tracks the worldwide theatrical box office. This is a useful signal, but it doesn’t capture all of a film’s income, and over the period I studied, the industry's economics have shifted.

That said, given that my numbers are comparative rather than trying to be comprehensive, I don’t feel that either effect is skewing the overall findings.

For each film, I built a relative fame score for its lead actor from a blend of publicly visible popularity signals gathered across several sources, averaged over the year ending one month before release, so the film’s own launch could not inflate it. Each film took the score of its most famous lead, and films were ranked against others from the same five-year window, so ‘top 10%’ means famous for its time rather than famous today. Leads were also classified by whether that attention was rising, steady or falling over the same window.

The median ratio is used throughout because the average is skewed by the occasional billion-dollar hit. Some genre and tier cells are small, in particular top-tier horror (10 films) and romance (22 films), so those specific figures are best read as indicative. Stars are also cast into better-financed films in the first place, so part of any apparent star effect is really a selection effect, which is exactly why the franchise comparison matters.

Last month, YouTube took offence to one of my Film Data Scientist videos on the decline in sex on movies. I got it up, but YouTube couldn’t last 10 minutes, and the video went down.

But, like a jilted lover, I’m back. I have recut the video to make it a little less sexy, and (so far) YouTube seems to be okay with it.

So here’s my explanation of why sex is disappearing from modern movies:

If the video has made you want to explore the topic a little deeper, here are some past research pieces I’ve conducted:

• Why is sex in movies declining? This examines long-term trends in sexual content in mainstream cinema and examines how global markets, franchise economics, and changing audience incentives affect what gets financed.

• Are erotic thrillers dying out? An analysis of the rise and fall of the erotic thriller, including production volume, box office performance, and the industrial conditions that once supported the genre.

• Which countries care most about sex in movies? A comparison of international ratings systems and cultural responses to sexual content, and how these influence distribution and revenue potential.

• What is replacing love as the dominant theme of movies? An exploration of shifting narrative motivations in mainstream cinema, and how romantic and sexual storytelling has lost ground to other emotional drivers.

• Sexual violence in mainstream movies. A focused look at how sexual violence appears in commercial films, and how its treatment differs from consensual sexual content.

• The sex, drugs and violence contained in MPAA ratings. A breakdown of how the US ratings board classifies sex versus violence, and what that means for studios trying to maximise audience reach.

I’ll be uploading new video essays every couple of weeks throughout 2026, so make sure you’re subscribed on YouTube to get notified when another drops (also, subscribing will really help me out with the Algorithm Gods, so it would mean a lot to me - thank you!).

https://www.youtube.com/@TheFilmDataScientist

And to help explain the channel, I also created a short channel trailer:

I have spent the past week talking with people, walking the Marché, dropping in on panels, and listening. My overall impression was that more is shifting under the surface than the official press releases reveal.

While Cannes 2026 felt a little calmer than last year, it wasn’t for reasons worth celebrating. Film professionals felt more focused, resigned to some changes, oblivious to others, and parts of the industry seemed thinner around the edges.

Here are the nine things I took away from this year’s festival:

1. Cannes got bigger and quieter at the same time. Record attendance, thinner buzz, fewer parties, shorter stays.

2. The word “strategic” was everywhere. Fewer meetings, more deliberate ones. Risk mitigation dressed up in a less scary-sounding word.

3. AI is regarded as an analyst, not an author. And unlike previous tech, it is not going anywhere.

4. Hollywood and streamers were nowhere. No studio tentpoles, and some key streamers felt less relevant in conversation and the press.

5. Japan was everywhere. Country of Honour, attendance up around 50% and a new Japan IP Market.

6. Private money and co-productions are the talk of the town. As pre‑sales have weakened, private equity has stepped in.

7. Europe is organising to protect itself. 4,700 film professionals signed a letter to the EU, Mubi launched a co‑financing pact, German auteurs revived Dogma.

8. The market is still trying to understand the kids. Gen Z is the only growth story, and the market is still deciding how to adapt.

9. And the Marché is trying to understand TikTok. First Creator Economy Summit, the largest virtual production stage at any film market, an AI for Talent Summit in its second year.

Let’s go through each in turn.

1. Cannes got bigger and quieter at the same time

By the headline numbers, this was the biggest Marché yet. Reuters and Variety reported 16,000 accredited participants from over 140 countries, up from 15,000+ in 2025. Deadline counted 1,700 buyers, 600 exhibiting companies, 250 industry events, and a festival‑wide footprint of around 40,000 professionals.

And yet the energy was thinner. I’ve noted in the past how an increasing number of industry veterans are choosing to skip expensive booths and passes in favour of just meeting in hotels and cafés. 2026 only seemed to accelerate this, with the Palais fighting a losing battle to remain the centre of the market.

Stays were shorter, schedules tighter, and the social calendar continued its decade‑long shrinkage with far fewer parties than ten years ago.

As the pandemic began to ease, many asked what would become of in-person events such as Cannes. I think we‘re starting to see the answer emerge. The time‑consuming bits of dealmaking can be handled by email, Zoom and screeners, so Cannes is where humans confirm in person what has mostly already been agreed.

So while I don’t doubt the 16,000 attendance figure, I suspect it doesn't mean the market is healthier, but just that the market has more spectators.

2. The word “strategic” was everywhere

If I had to pick one word that has run through my conversations this year, it would be "strategic." I heard it used by buyers, sellers, financiers, even the festival programmers used it, when describing why they had selected the films they had.

What people mean by it:

• Fewer meetings, but more deliberate ones.

• Fewer films, but more carefully packaged.

• Fewer parties, but the right ones.

• A market that has lost its appetite for discovery and instead follows a narrow checklist.

Roeg Sutherland, co‑head of CAA Media Finance, framed it as an opportunity, telling The Ankler:

The market is absolutely disrupted, but in that I always see opportunities. There are opportunities to build distribution companies. There are opportunities to make movies for the right price.

Fionnuala Jamison, managing director of mk2 Films, framed the new buyer brief in a Screen Podcast interview thus:

What we’re hearing from buyers is that they want the best films, great filmmakers on top of their form. Then we have to look at how difficult the subject is potentially and how demanding it might be on the audience in terms of filmmaking style. Then it’s factors like uniqueness, pitch, and whether there is an obvious marketing hook for distributors.

This is risk mitigation by another name, and it tracks with what I picked up at the EFM in February. The default setting for everyone in the room is now caution… even if they don’t use the word.

3. AI is regarded as an analyst, not an author

In Cannes 2025, it felt like AI was everywhere, but nobody knew what to do with it.

In 2026, it seems that people are comfortable saying out loud that AI is being used to analyse projects, such as simulating focus groups, profiling audience overlap, generating storyboards, and suggesting marketing copy, and panels openly discussed how AI agents can provide you with an “unfair advantage”.

Just an aside, I have genuinely lost count of how many people have pitched me their new start-up, which can turn your script into a storyboard or do pre-production work. I was trying to keep count for this article, but I foolishly didn’t write them down, and now all I can say is that it was more than twelve.

What remains unsayable in polite company is AI as creator. Steven Soderbergh’s John Lennon: The Last Interview, which uses Meta‑generated video of Lennon and Yoko Ono for around 10% of its runtime, sparked a lot of debate. Variety reported that Patrick Wachsberger’s Legendary‑backed sales company 193 has boarded the film, with Meta listed as the film’s technology partner. The Meta partnership at the festival itself, including the “Meta House” at the Majestic and Ray‑Ban smart glasses on a brigade of influencers, prompted a backlash captured by nss magazine.

Weeks before the festival opened, around 4,000 French actors signed an open letter in Le Parisien, including Bérénice Bejo, Léa Drucker and Swann Arlaud, describing AI as a “devouring hydra” engaged in “organised plundering”.

Demi Moore told Variety:

AI is here. To oppose it is to engage in a battle we are likely to lose.

Peter Jackson, accepting an honorary Palme d’Or, told The Hollywood Reporter AI was”

…just a tool like any other…no different from other special effects”.

The strangest case study sits just outside the festival. Gossip Goblin, the kitchen‑table AI filmmaking outfit run by Zack London from a Stockholm apartment with a team of eight, has clocked over 500 million views on its short satirical AI films. The Hollywood Reporter reported that LA agents, producers and studios are now flying to Stockholm to meet him (note that they are not coming to Cannes to pitch their wares).

Cannes often sees technologies come and go (3D, VR, NFTs, Metaverse, blockchain, etc.), following the same hype-to-bust playbook. But AI is not like others

I tracked mentions of new technologies in the Cannes trade‑press daily from 2019 to 2026 (Screen Daily, The Hollywood Reporter and Variety) and noted how often each technology was mentioned (my methodology is at the end of this article).

This reveals that AI is not just a new fad, neither in scale nor fleeting nature.

It’s worth noting that not all the ‘blip’ new techs are the same. While NFTs and VR are genuinely absent, blockchain technology is powering many of the startups in Cannes. It’s that the people have learned not to use certain words, and instead focus on the benefits, not the underlying technology.

4. Hollywood and streamers were nowhere

For the first time since 2017, no major Hollywood studio brought a tentpole to Cannes. In another life, we might have expected to see Cannes premiere Spielberg’s Disclosure Day, Nolan’s The Odyssey, the Cruise/Iñárritu Digger.

Read more

No one says "Follow that car!" anymore... and there's a reason why.

I ran the numbers on over 72,000 films to find the most reused lines in cinema, revealing how movie clichés are born, spread, and die.

Chapters

• 00:00 - Movie clichés are good, actually

• 00:50 - The way you speak sucks

• 03:40 - Movies are talking to each other

• 04:57 - Bored yet?

• 08:09 - Word Salad

• 11:39 - Deja View

Further Reading

If you want to explore the research behind this episode in more detail:

• What are the most commonly used movie clichés? I ranked the most reused lines in mainstream cinema, with breakdowns by genre, decade and frequency.

• Has the way movies use dialogue changed? I looked at how line length, dialogue density and pacing have shifted over the decades, and what that tells us about how modern films deliver story through speech.

• Which genres dominate when movies blend styles? My study of how genres sound different from each other, including which films talk the most, which leave the longest silences, and how dialogue patterns travel when films mix genres.

I’ll be uploading new video essays every couple of weeks throughout 2026, so make sure you’re subscribed on YouTube to get notified when another drops (also, subscribing will really help me out with the Algorithm Gods, so it would mean a lot to me - thank you!).

If you want to know more about the channel, here’s a quick primer:

Winning a Palme will certainly give you a moment of fame, the adulation of filmmakers, and a chance to boost your next project. But will it help your movie’s chance at the box office?

All films which are ‘In Competition’ at Cannes have not yet been released to the public anywhere in the world. Some may have a release lined up, but most will deliberately leave negotiations open, just in case they can use the Cannes buzz to get a better deal.

But does winning help?

We can’t definitively prove cause and effect, but we can certainly look for patterns in the data. I built a dataset of every main competition film at Cannes, Berlin and Venice between 1990 and 2024. That gave me 2,159 competition slots across 35 years.

Do Cannes-winning films perform well at the box office?

Let’s start with the headline figure - yes, films which win the Palme d’Or do significantly outgross their fellow nominated films in commercial cinemas.

Cannes top-prize winners typically take roughly nine times the worldwide gross of the other films competing alongside them.

Berlin and Venice top winners do better than the rest of their slate, too, but the gaps are far smaller and not stable enough to build a business plan on.

Cannes has some standout winners:

• Pulp Fiction took $213m worldwide on the back of its 1994 Palme.

• Parasite took $254m in 2019.

• Fahrenheit 9/11 took $221m in 2004.

• The Pianist took $112m in 2002.

• Anora took $58m last year .

• Anatomy of a Fall $49m the year before.

Berlin’s top of the pile contains:

• Spirited Away ($396m)

• Sense and Sensibility ($135m)

• A Separation ($24m)

Venice has had three blockbuster top winners in the last seven years alone with:

• Joker ($1.1bn)

• The Shape of Water ($196m)

• Poor Things ($112m)

So the difference with Cannes winners is almost nine times and still a respectable 2.8 times at Berlin, and around 1.5 times at Venice. These three festivals are not the same!

But is it really the Palme doing the work?

There are two things we should take into account when considering this finding: quality and language.

Firstly, a film that wins the Palme is probably a better film, on average, than one that doesn’t. When we crunch the numbers, it turns out that, yes, festival juries at all three festivals select films that critics and audiences rate above average.

The second factor we need to consider is language. It’s worth noting that the Palme has tilted noticeably toward films with English-language distribution partners in recent years. Anora, Parasite and Triangle of Sadness all had US distributors locked in early.

To check whether the prize itself is doing the work, I compared like with like. I held the film's language constant first (English-language films earn much more, regardless of any prize), then held the IMDb rating constant on top of that (better films earn more, regardless of any prize).

So what we’re left with is data on how big the Palme bump is at each stage of stripping out those two factors. By the time you’re comparing Palme winners against other Cannes competition films with the same language and a similar rating, the bump is down to a bit under three times the typical gross.

That residual is the part the prize itself can plausibly take credit for; the rest of the headline gap is the prize tracking films that were already commercially advantaged.

When I did the same maths with the winners at Berlin and Venice, I found nothing survived. Once you compare like with like, neither festival’s top prize lifts a film above the rest of its competition in any way that the data can detect.

None of this means the prizes are worth nothing. Career legitimacy, future financing, critical standing, distributor curiosity, the right to put a laurel on a poster for the rest of your life.

But the specific claim, that winning a top festival prize equals a commercial bump, turns out to be a Cannes-specific claim, and less than it initially seems.

Bonus - Is the Cannes winner usually the best or biggest film of its year?

As a side note to finish on, I also looked into whether, when a jury awards a Palme, Lion or Bear, they are handing it to the highest-grossing or best-rated film in their year’s nominations.

For example, Parasite won the Palme d’Or and became the highest-rated film on IMDb in the entire 2019 Cannes competition. But it was not the highest-earning hit of the cohort, being beaten at the worldwide box office by Once Upon a Time in Hollywood ($377m versus Parasite's $254m).

It turns out that Berlin and Venice juries find the top-rated films about a quarter of the time (not bad for a lineup of 20 to pick from), but Cannes lags behind at just 14%.

The top-prize winners are rarely the highest-grossing films in the cohort.

If you want to guess if a Cannes-nominated film will do well in cinemas with the public, just look for famous names. Or Shrek. Mostly Shrek.

Notes

Today’s analysis covers every main competition film at Cannes, Berlin and Venice between 1990 and 2024. Sidebar sections such as Un Certain Regard, Forum, Orizzonti and the Director’s Fortnight are not included.

I used 1990 as the cutoff because earlier years have patchier box office records and many older films’ totals are inflated by re-releases that no modern release would benefit from, which distorts comparisons.

To prepare, I have gone through the conference schedules for this year (as well as 2024 and 2025) to see what these 742 events reveal about the film business in 2026.

I’ll go through it in detail later in the article, but here are the key themes as I see them:

1. The conference is shrinking (slightly). The 2026 edition has fewer sessions and about 16% fewer programming hours compared to recent years, despite an increase in attendance.

2. AI is fully integrated. 2024 was about whether AI mattered, whereas 2026 is about what AI agents do for sales teams, how to license a synthetic voice, and how to plug an AI compositor into a virtual-production pipeline.

3. The streamers have stopped showing up. Netflix, Amazon, Apple and Disney have gone from nine speaker billings in 2024 to one in 2026. Whatever Cannes is for now, it isn’t appealing to streamers.

4. Co-production is the survival mode. When financing is tight and audiences are fragmented, splitting the budget across borders is how you keep working. 32 sessions in 2026 are explicitly about it.

5. Virtual production is going hands-on. It launched last year as a panel strand, but this year converted 13 of 17 sessions into workshops. The conversation has moved from “should we?” to “how do we?”.

6. The values agenda has thinned. Diversity and inclusion content has nearly halved. Sustainability is one event in 243. Mental health is two. Whatever the industry is talking about in 2026, it isn’t the values conversation it was having in 2024.

7. Creators are being grafted onto cinema… cautiously. The brand-new Creator Economy Summit on 17 May is six panels in one day with Scalable, Meta and YouTube. Cinema is now talking to TikTok-native creators on cinema’s terms, not theirs.

Let’s look at each in turn…

1. The conference is shrinking (slightly)

Despite the festival being the world's largest film market, its conference offerings are shrinking.

In 2024 the Marché ran 260 conference events totalling 379 hours of programming, but this year that’s dropped to 243 events and 318 hours. The mean event length has also shrunk, from 88 minutes in 2024 to 79 minutes in 2026. The schedule is doing fewer things, more briefly.

The Marché itself reports growing numbers of accreditations, and the side-strands keep multiplying so this isn’t a lack of willing attendees. Rather, it’s a curatorial set of decisions that has meant the schedule has been trimmed year after year.

Another such decision is that Cannes 2026 seems to have converged on a single format. 48% of all conference events in Cannes are multi-speaker panels. Panels are the only format that has grown across all three years. They went from 95 in 2024 to 116 in 2026, a 22% rise. Almost every other format is flat or shrinking.

My theory is that it is a mix of programming-equity question (panels can include four sponsors at once), and partly a product of market forces (a sponsor will pay to have its CEO on a panel, whereas it is much harder to sell a slot in someone else’s keynote).

So if you’re hoping to hear a single experienced voice speak at length about their craft, the Marché might disappoint. But if you want six middle managers from different companies offering carefully coordinated takes on what AI means for distribution, then you’re in for a treat.

2. AI is fully integrated

27 events mentioned AI in 2024, 21 in 2025, 31 in 2026. The 2025 dip looks like fatigue after the immediate post-ChatGPT bump. In 2026 it’s back, stronger, and at a higher share of the total schedule.

The framing of AI has also shifted. 2024 panels were “AI for filmmakers, what does it mean”, “Curious Refuge: case studies in AI cinema”. By 2026 the sessions are more like middle manager meetings:

• How AI Agents Give Top Sales Teams an Unfair Advantage

• AI Meets IP: Voice, Rights and Creative Control

• From Legendary Studios to AI Pipelines in Film Music

There is even an “AI for Talent Summit” running across two invitation-only sessions on Friday 15 and Saturday 16 May. AI in casting agencies is now its own conference-within-a-conference.

Two things missing from the 2026 AI line-up are notable. Neither OpenAI nor Anthropic has any speaker presence, despite 31 AI-related sessions. The companies on stage are AI-tooling startups serving film (Phonoma, Respeecher, Eros Innovation, Largo, Wonder Dynamics, Dreamina AI). The frontier-AI companies that drive the underlying conversation are not there. Make of that what you will.

3. The streamers have stopped showing up

Looking at who’s speaking, we get an interesting story of change. The three areas in decline are:

1. Festivals (39 in 2024, 24 in 2025, and 14 this year)

2. Streaming platforms (9 → 6 → 1)

3. Trade press (26 → 15 → 9)

Read more

Predicting who will be the next James Bond has become one of the great communal conversations. Whether that’s via prediction markets, professional bookmakers, newspaper headlines, or just down the pub - everyone has an opinion.

I think it should be said loud and often that this is a deeply silly thing to bet on. Predicting the Bond casting is less of a science, more like forecasting the weather next Christmas (ie, will it be white).

There are real underlying patterns in the historical data, but the actual decision is being made in private by a small group of people working with information you and I do not have. Even the professional bookies are essentially trading on rumour, public sightings, agent gossip, and the occasional carefully-placed leak.

I hope the silliness of my video above conveys how seriously this should be taken. BUT… that’s not to say the topic lacks merit. A number of important ideas around casting, acting, and IP management emerge as we frolic through the data. So, let’s take a look with a view to gambol rather than gamble.

The current state of play

At CinemaCon 2026 a few weeks ago, Courtenay Valenti, Amazon MGM’s head of theatrical said:

Now, I know you’re all wondering when we’re going to announce who’s playing James Bond. Please know that we’re taking the time to do this with care and deep respect. It is the dream of a lifetime for all of us to bring audiences this next chapter, and it’s a responsibility we don’t take lightly.

What I can tell you is this: when you pair one of the most beloved franchises in history with a world-class filmmaking team… setting the stage for something that’s truly worthy of the Bond legacy. That film is coming, and when the time is right, we’ll have much more to share.

This means we all have a brief window in which to have fun speculating!

Let’s put ourselves in the shoes of the top Bond brass at Amazon. You have been handed the most consequential casting decision in commercial cinema. You don’t have a name in mind. You have a brief, a budget, a multi-billion-pound franchise, and a producer breathing down your neck. How do you build the shortlist?

The honest answer is that it is a series of cuts. Some hard. Some soft. Some defensible by data, others by gut feel and politics. By the end of the process, you should have a list of perhaps five names sitting on a desk in a building Amazon hasn’t shared with the rest of us.

Let’s see how far we can get with the public info we have...

The hard requirements

The hard requirements are the cuts you can defend with a sentence. Each one is a decision a producer would actually make in week one of the search.

1. Male

Bond producer Barbara Broccoli said only a couple of years ago that:

He can be of any colour, but he is male.

It’s not clear to what extent Amazon agrees, but the vast majority of the public does. A YouGov poll from 2023 asked Americans which traits a future Bond must have.

The public is conservative on gender (72% want a man) but genuinely open on race (only 39% require a white actor).

So when doing my first pass cut of names, I followed the majority on the male requirement, and left race open.

2. British (or convincingly so)

This is the cut where the data and the principle disagree.

The whole concept of acting is one about convincing people you’re something you’re not. Therefore, on the one hand, it seems absurd to place a birthplace requirement on the person who goes running around with a fake gun, fake-killing other liars in a fictional story we all know didn’t happen.

But now we touch on another aspect of casting - believability. The audience needs to be willing to suspend their disbelief and stop seeing the actor and start seeing a real superspy. And nationality plays a part in this shared public delusion.

Deadline reported in September 2025 that the casting brief for Bond 26 is:

British, male, unknown, late 20s to early 30s.

So whatever I think about national-flag accents, it seems like I need to keep it as a hard rule with which to filter our candidates.

(The rule I settled on was more accurately “The British Isles” than just “Great Britain”. The differences between all the ways people describe the islands is a massive rabbit hole. If you have two minutes, watch this; if you have two hours, watch this).

3. In their thirties

Amazon has to be hoping that they pick the right person now and get to ride that successful casting for at least a decade, maybe two. That places extra pressure on the person to be as young as possible, while still pulling off the role in their debut Bond (likely release date is 2028)

Broccoli told Variety in 2022

When we cast Bond, it’s a 10, 12-year commitment. So he’s probably thinking, do I really want that thing? Not everybody wants to do that.

The two most commercially successful Bonds were the youngest at debut. Connery had a nine-year run from 32. Craig ran from 38 to 53 across five films and $3.97 billion in box office. Moore made seven films but was visibly ageing out by the end.

4. A working professional with real films behind them

Playing Bond is a hard gig, both on- and off-screen. Long-press tours, multi-year shoots, action choreography, and the requirement to anchor a film against an A-list villain.

The clearest description of what this means in practice comes from Debbie McWilliams, the casting director who has been with the franchise since Roger Moore and personally cast Dalton, Brosnan and Craig. She told SlashFilm:

He has to look like a regular guy. You can’t be Dwayne Johnson. He has to have a great physique, but he shouldn’t stand out in any situation. You feel a very strong presence in the room with him, and I think that is incredibly important.

In my filtering, I opted to look at people with at least three feature film credits, although obviously, a real casting director can take a more nuanced view on who is able to take on the mantle.

The soft requirements

Once we apply these filters, there are fewer than 500 candidates left. We’re still at the point where we couldn’t call them all in for a casting session, but we could start a spreadsheet.

The hard requirements give us a list of professional actors, and now we can use soft requirements to surface people we might otherwise overlook.

Although data can give us pretty and definitive-looking charts and tables, we are firmly in the ‘Cosplaying as a Casting Director’ territory at this point. It’s a fun thing to make YouTube content with, but let’s not confuse this with hard science.

For example, acting ability is important here. Bond isn’t a dumb 80s action thriller where we can overlook the leads' acting because of the size of their quads or how often they can do the splits. A real casting director will know in seconds whether the actor is the worst thing in a good film or the best thing in a directionless one. But the data can’t see this kind of thing.

So let’s see what we can measure to reorder the candidates in interesting ways.

1. Has prestige

The actor who takes on Bond will need to have a certain amount of heavyweight prestige about them. They can’t be a silly Adam Sandler / Kevin James type.

Looking at the Metascore of the movies each of our Bond-wannabes has been in, surfaces Paul Mescal, Joseph Quinn, and Josh O’Connor.

When asked a couple of years ago about the role, Joseph Quinn said:

Yes, in fact I just did a Zoom with Barbara… It would be fucking stupid to say no that… But come on, it doesn’t even make sense to think about it.

2. Doesn’t often feel miscast

Another signal we might look at is how well the actor has convinced the public in their previous roles. I pulled audience reviews and tracked mentions of “miscast”, “wrong for the role”, “unconvincing”, “didn’t fit” with each actor.

This could just be a case of not being cast correctly in the past, or it could be a signal that audiences are unsure about them.

An actor flagged as miscast might be miscast for a specific film because the casting director picked the wrong vehicle. Or because the director did not get a performance out of them. Or because they have not yet found their niche.

Even with all those caveats, the fact that Callum Turner, the current bookies’ favourite, sits in the worst third of his peer group on miscast mentions might be a concern.

3. Lots of charisma

The third proxy is the strongest of the three because it is the one most directly about the actor. Using the same audience review corpus, I tracked counts of mentions of the actor alongside things like “charisma”, “magnetic”, “presence”, “star quality”, “commanding”, and similar euphemisms.

Jonathan Bailey and Aaron Talyor-Johnson came up top (with the latter having a long history of being linked to the Bond gig).

4. The right kind of famous

The perfect actor to play James Bond needs just the right amount of fame upon casting. Bond does not hire stars. Bond makes stars. McWilliams calls casting an unknown “a gift”.

For most lead roles, more fame is better as fame brings audiences, marketing, and secured financing. For Bond, fame above a threshold is a negative. They do need some basic level of fame so that audiences feel they have earned the role with a sense of gravitas (gone are the days of hiring a milkman to play Bond).

They also need to be affordable and controllable - two things that famous people are not. A famous star walks into the negotiation with leverage Amazon cannot compete with.

Bond casting is structurally closer to casting Harry Potter or Star Wars, in that you are not picking the actor someone is right now, you are picking the actor they are going to become.

For example, Tom Holland is a perfectly plausible Bond on every other criterion (except maybe height). He is also already in a Marvel contract that would constrain availability, already a household name, and already on his own profit participation deals. His audience is not Bond’s audience either. The studio loses leverage the moment they sit down with him.

The Daniel Craig Precedent (a.k.a. the Renée Zellweger Principle)

When Daniel Craig was announced on 14 October 2005, the internet had something close to a meltdown. He was blonde. He was 5’10”. His ears were “too prominent”. Two protest websites were launched within hours. Tabloids published photos of him in a life vest, struggling to get out of a speedboat. Den of Geek covered the backlash in detail.

Then Casino Royale opened to 94% on Rotten Tomatoes. Craig’s five films grossed $3.97 billion. Skyfall became the first Bond film to cross a billion dollars.

The same thing happened five years earlier with a much less British example. When Renée Zellweger was cast as Bridget Jones in 2000, the British fanbase was furious. A Texan playing a national icon. Zellweger gained weight, mastered the accent, worked undercover in a London publishing house, and won the audience over. We do not now think of Bridget as miscast - we think of her as Bridget.

The rule has held for sixty years - Connery was not Fleming’s choice, Lazenby was not anyone’s choice, and Craig was too blond. You could argue that the audience does not know what it wants until you show it. And every time the producers have made a casting choice that contradicted public expectation, the franchise has thrived.

The $2.8 trillion elephant in the room

This is the part of the problem that the data cannot solve. The casting decision lies at the intersection of two shopping lists. The audience wants their Bond. Amazon wants their franchise. Those are not the same thing.

• Audiences want a Bond they recognise the moment he walks on screen, someone who feels like a continuation of the canon rather than a hard reset. He has to look the part, with the kind of bearing and physical presence the role demands, and light up a room the second he enters.

• Amazon wants someone cheap to sign, because the franchise is the star and the actor is replaceable. They want someone willing to commit to a TV universe of spin-offs, not just two films and out. They want an actor who’ll tolerate director-led work without demanding the final cut, and who’ll stay available and under contract for the next decade.

• The person who satisfies both columns is likely to be British, male, in his late twenties or early thirties, charismatic and properly trained, a working actor who isn’t yet a star, and someone who takes direction rather than fights it.

The true overlap among those people is no more than a handful. A source close to production told Deadline last year that:

99.9% of the names speculated online so far won’t make the cut.

Gun to the head….

I think it would be rather unsatisfying if I ended this article without proffering some names I think could be Bond.

Rather than making a subjective or artistic judgment, I’ll tell you what the numbers say.

At the time I was researching the video at the top of this article, the frontrunner was Callum Turner, with odds which imply he had already been cast. In my data, he scored pretty poorly. He has the third-highest miscast rate in the famous shortlist, and sits near the upper end of the casting age window.

If I had to pick someone who scores well but is priced badly by the bookies, it would be Tom Glynn-Carney. 31 years old, six foot one, a graduate of Guildhall, average film Metascore 65.2 and with a Wikipedia footprint of a quarter of a million pageviews (so known but a tenth of Tom Holland’s numbers). He has the shape of every Broccoli-era casting decision the producers have made.

Other top choices the data liked were Fionn Whitehead, Joseph Quinn, and Nabhaan Rizwan.

But any tweak to the criteria would massively change the order. Reweight charisma by even a few percentage points and Jonathan Bailey or Taylor-Johnson is on top. Decide that fame is a positive signal rather than a negative one, and Holland leads. Drop the British requirement, and Jacob Elordi is your Bond.

Bond casting is not really a maths problem. The producers are looking for the right unformed clay who walks into the room with the right ingredients, and who they can write a script for, build a film around, and brand for a decade. Oh, and who they can afford and control.

While data gives us fun candidates to consider, this kind of decision is only ever made by a small number of people over coffee in a private room.

Andrew Sarris, in his 1963 essay “The Auteur Theory and the Perils of Pauline” reflected on the emerging idea of a film auteur like this:

After a given number of films, a pattern is established, and we can speak of the Rays, of Ophuls, Renoir, Mizoguchi, Hitchcock, Chaplin, Ford, Welles, Dreyer, Rossellini, Murnau, Griffith, Sternberg, Eisenstein, Stroheim, Bunuel, Bresson, Hawks, Lang, Flaherty, Vigo, as we speak of artists and authors in other media.

Thirteen years later, Hitchcock put it more pithily:

Self-plagiarism is style.

Both were saying the same thing from opposite ends of the room. Sarris saw the pattern across a director’s films as what earns them the title auteur. Hitchcock saw the same pattern as a method, chosen on purpose.

Either way, the claim is that a real director leaves a visible mark across the films they make. I wanted to see whether that was actually true.

So I took 2,970 live-action features, directed between them by 190 different people, and for each director, I measured how similar their films were to one another. Every director had to have at least four features in the sample, because below that, the score gets noisy and tells you more about the individual film than about the director. There are more technical details in the Notes section at the end.

Just for the avoidance of doubt, I’m not implying that being visually consistent is either good or bad. Sometimes a director is making very different films on purpose, and a new look is part of the point. Sometimes a director refines one look over a career, and the repetition is the point.

The directors whose films look most like each other

I used image embeddings of all shots in the movies to generate a Consistency Score. This is a single value per director which reveals how much more alike a director's films look than a random batch of films from the same genres and decades.

• A score of zero means the films are no more or less alike than a random batch.

• A high positive means they cluster tightly together.

• A negative means they scatter even more than the baseline.

At the top sits Taylor Hackford with a consistency score of +0.38.

Below him, the list mixes directors who lean into a very specific genre lane (horror, family comedy, franchise action) with a couple of names more familiar to cinephiles. Alfred Hitchcock sits at position 7. He is the only director in the top ten who is routinely called an auteur in the traditional sense.

Hitchcock storyboarded every scene in advance and arrived on set, treating filming as the execution of a finished plan so it iis not surprising that his films shared a visual language.

What happens when a director changes the cinematographer?

But directors do not work alone. The film’s final visual style is a collaboration with the cinematographer (also known as Director of Photography, or simply “DP”).

So I wanted to see what happens to the numbers when we narrow the list to directors whose films span at least three different cinematographers. The top of that filter is the same core names as before, because several of them were already three-plus-DP directors, but we lose Hitch.

The directors who reinvent their look for each film

At the other end of the ranking sits Mel Gibson, followed closely by Nimród Antal and Vincenzo Natali.

Most of the names in the bottom cohort are not the ones a reader reaches for first when asked to picture an auteur. Some are genre specialists working across horror, thrillers and action (Alexandre Aja, Jaume Collet-Serra, John McTiernan). Others are career directors whose filmographies happen to span wildly different territory (Brian De Palma, Niki Caro, David O. Russell).

What they share, by this measure, is that the films they actually made look strikingly unlike each other.

The directors who work this way are often explicit about why. Danny Boyle has made the case for reinvention as a creative discipline:

Always changing genres, making very different films is a good idea. It’s a way of making yourself feel vulnerable again, getting back to that innocence.

Steven Spielberg said of Stanley Kubrick:

He kept saying, I want to change the form. I want to make a movie that changes the form. And I said, Well, didn’t you do that with 2001? He said, Just a little bit, but not enough. I really want to change the form.

Consistency is not the same as distinctiveness

Our metric so far is good for answering the question ‘Does this director’s filmography repeat itself?’, but we can use the same underlying data to also answer another question. Namely, ‘Does this director look unlike other directors?’

A director’s filmography can be tight and generic, or, conversely, varied and unmistakable.

So I calculated a Distinctiveness Score, which measures how far a director’s average film sits from the nearest other director’s average film in the same measurement space. Higher means their work sits further away from everyone else’s.

If we plot both metrics on the same chart, we get to see four quadrants:

1. Consistent and distinctive (top right). The classic auteur zone. Films that look like each other, and like nothing else in cinema. Repetition that adds up to a signature.

2. Varied but distinctive (top left). Every film is a reinvention, yet every film still sits somewhere unusual on the map. The signature is in the ambition rather than in any single composition. You cannot mistake the filmography for anyone else’s, even if you cannot mistake one film for another either.

3. Consistent but generic (bottom right). Films that look like each other, but also like plenty of other directors’ films. Usually, a genre specialist working inside a narrow lane, where the commercial logic of the genre is doing as much work as the director is.

4. Varied and generic (bottom left). No through-line, no signature. Working directors taking whatever script the week brought them, inside a part of the map that is already well-populated.

This shows us that two commonly cited aspects of the auteur label (i.e., a director with a visual style and a director who presents unique work) don’t always go hand in hand.

In future research, I’m going to quantify the extent to which the key creatives affect the final film’s visual style.

Notes

For every director with four or more live-action films in the sample, I computed the average similarity between the films’ image fingerprints, then subtracted the average similarity of a matched random sample of films from the same genres and decades.

• A positive score means the director’s films hang together more tightly than a random mix of the same genre and decade.

• A negative score means the opposite.

• Near zero means a filmography is no more or less consistent than the matched baseline.

In the sample were 190 directors, 260 cinematographers and 2,970 live-action features. Documentaries and animation are excluded. Directors or cinematographers whose careers are primarily in animation are dropped, since animated films are a very different beast.

Each person needs at least four films in the sample to appear on the leaderboards; directors below that threshold are excluded because their scores would be noisier than meaningful.

Each film is represented by a summary fingerprint derived from stills taken at every detected scene boundary. The fingerprint captures overall composition, subject and palette, enough to tell apart a Wes Anderson frame from a Michael Bay frame at a glance, though not enough to know which lens was on the camera. Cosine similarity between two fingerprints gives a single number for how alike two films look.

When a director works with one cinematographer across every sampled film, their consistency score also carries the DP’s fingerprint. The three-plus-DPs chart is there to surface that.

Directors with three or more films inside a single franchise (for example Harry Potter or John Wick) score higher because the franchise brand enforces a look.

The measure captures composition, not dialogue, sound, music, editing rhythm or thematic preoccupation. Several directors whose signatures lie outside the frame, such as Lynch in texture and sound, or Tarantino in dialogue and shot-type idioms, appear lower than their reputations suggest.

For the scatter chart, each director has a centroid in a standardised feature space built from sixteen composition metrics. Distinctiveness is the distance from that centroid to the nearest other director’s centroid. Higher means further from everyone else.

Cannes Confidential

6.61MB ∙ PDF file

Download

The Raindance Guide to Navigating the Cannes Film Festival and Marché du Film

Download

Download

Every spring, around this time, my inbox develops a particular shape.

Somebody emails to ask which badge they need. Somebody else emails to say they’ve been told Cannes tickets are ‘a lottery’ and is that really true? A first-time producer wants to know if they should bring business cards or a link. A publicist is panicking about the new dress code. And at least one person, every single year, asks me what shoes to wear.

The 79th Festival de Cannes opens on 12 May 2026, with the Marché du Film running 12 to 20 May. Around 15,000 market professionals from 140 countries will be there, and at least a third of them will be first-timers.

I’ve pulled together the questions I actually get asked, the ones people don’t ask until they land, and the ones they should ask but don’t. What follows draws on a survey I ran of 575 Cannes veterans, the festival’s own documentation, and the 2025 and 2026 rule changes that have caught several return visitors off guard.

Accreditation

Which badge do I need?

Six tracks, but most first-timers only need to worry about two.

If you’re a working industry professional who wants access to the Palais, Official Selection screenings, pavilions and panels, you want Festival accreditation. It is ‘free’, with an asterisk. A mandatory €24 (inc. VAT) environmental contribution, introduced in 2021, is non-refundable even if your application is rejected.

If you are a producer, distributor, sales agent or financier who wants access to the market, its screenings, Cinando and the Marché events, you want the Marché du Film badge. That is paid, and not cheap. Screen Daily’s 2024 reporting put it at €429 early-bird, €499 standard, €600 late. Marché director Guillaume Esmiol told Screen Daily that about 40% of badges go at the early-bird rate and only 1% at the late walk-up price.

The other four tracks: Press accreditation (tightly policed, returning journalists must submit prior coverage), 3 Days in Cannes (free, for cinephiles aged 18 to 28), Cinéphiles (for locals, via the mayor’s office), and SFC Rendez-vous Industry — formerly Short Film Corner — for short filmmakers with a project in the Market Catalogue.

One counter-intuitive fact: actors are not eligible for a Marché badge. They can only apply for a Festival badge. Producers, distributors, sales agents, exhibitors, financiers and film-industry service companies are eligible for the Marché.

When does the portal open?

For 2026: accreditation opened on 2 February 2026 and closes on 15 April 2026 for standard applications. A late fee of €224 (inc. VAT) kicks in from 1 April. Multiple attendee accounts confirm that February and early-March applications for 3 Days in Cannes got in while late-March ones were rejected on capacity grounds — the “rolling review” line is real.

The badge colour thing: is the caste system real?

Yes. And it is brutal.

For press, the ranking runs white (top), pink with yellow dot, pink, blue, yellow (bottom). For industry attendees there are also grey, orange and black badges for crew, photographers and technical staff. Euronews described the hierarchy as ‘Divergent-style segregations’, which is the most accurate line anyone has written about it.

Your badge colour matters because it dictates when tickets appear on your dashboard each morning. The higher-tier badges get first pick at 7 am; lower tiers wait until 8 am; and on J-2 and J-1 — two days and one day before each screening — unclaimed tickets are released to all categories without distinction. That is not a minor detail. It is the single biggest determinant of which films you’ll actually see.

Tickets

Is Cannes ticketing really a lottery?

No.

The myth persists mainly because the outcome feels random. The official explanation, buried on the festival’s own ticketing page, is explicit: when the ticket office opens four days before each screening, quotas are initially allocated to each accredited category, with professional accreditations prioritised. One minute before seats are released each morning, users connected to the site are moved into a waiting room and admitted “based on a random draw”. That is the only genuinely random element. In 2025, 90% of accredited users logged in to the site in under 2 minutes.

So it’s not a lottery. It’s an algorithm rewarding badge status and early access.

How does the 4-day rolling window work?

For each screening, reservations open between four days and 15 minutes before the screening (one hour before for Grand Théâtre Lumière gala screenings). New seats appear at 7am or 8am depending on your badge tier. On J-2 and J-1, unclaimed inventory is released without category distinction. That’s the window where lower-tier and cinephile badges have their best shot.

In 2024, 95% of tickets were gone within 15 minutes of opening. In 2026, reservations open on 8 May.

Why do tickets show ‘sold out’ when the theatre is half-empty?

Three reasons. Attendees hoard tickets and no-show. Algorithmic quotas under-allocate to whichever tier actually turned up. And for gala screenings, big blocks of seats are held back for artistic teams and invited guests until the last minute.

Since the 2024 ticketing revamp, if you no-show twice you’re auto-blocked from bookings for 48 hours. Cancel at least an hour before the screening to avoid penalty. It’s rumoured that journalists are exempt from this rule, but I couldn’t get confirmation either way.

What do I do if I don’t get a ticket?

The last-minute queue is where underdogs thrive. In 2025, 18,000 seats were allocated through those last-minute queues. Competition films screen at the Cineum in Cannes La Bocca up to four days after their Lumière gala, and the festival reinstated its fourth large theatre there in 2025. The festival itself reports that ‘a huge majority’ of badge-holders queuing at the Cineum get in.

Official Selection films also screen an average of 3 more times after their première. So if you miss your must-see on day one, you haven’t missed it.

Any screenings that are easier to get into?

Yes, and the festival tells you this itself. First screenings around 8:30 am and last screenings from 10 pm are the least oversubscribed. Early birds and night owls eat well at Cannes.

The Marché floor

I’m a producer with a film to sell. Should I spend my week on the Marché floor?

Largely no, and this is the single biggest first-timer trap.

The Marché is enormous. Roughly 15,000 professionals from 140 countries, around 600 exhibitors, over 4,000 films and projects, 250 industry events, more than 80 keynotes, panels and conferences, and 1,500 screenings. The booth floor in the Palais basement — ‘the Riviera’ — is vast, loud, and optically impressive.

It is also where sellers are, not buyers. The buyers spend their time in hotel suites and balcony offices along the Croisette. That’s where acquisition executives take meetings.

Their day is a rotation of Cinando-booked calls, poster walls at the Majestic terrace, and the occasional screening. Very few of them walk the Marché floor in any serious way.

So what actually is Cinando, and why does everyone whisper about it?

Cinando is the year-round B2B database underneath the Marché, launched in 2003. Companies, contacts, projects, screeners, market attendees, and screening schedules. It’s where pre-festival meetings get arranged and where buyers, sellers and producers look each other up before the first espresso.

A one-year Cinando subscription is included with every Marché badge. This is, by some distance, the single biggest non-cash benefit of paying for the Marché tier. If you’ve paid for a Marché badge and you’re not using Cinando to pre-book meetings four to six weeks out (i.e. now!), you have effectively rented a car and left it in the car park.

What’s the Croisette Package, and do I need it?

For sales companies, studios and film institutions operating from hotels or apartments rather than taking a Palais stand, the Marché now requires the Croisette Package.

In 2024, it was priced from around €2,100 to €2,400 for three badges for companies with turnover under €2m, and €4,400 to €4,900 for six badges for companies above.

Individual Marché badges on their own no longer unlock the same off-Palais operating rights. Practitioners read it as a roughly €1,000 premium per company; the Marché defends it as sustainability- and sales-company-eligibility-driven.

What about the Producers Network?

The Producers Network is capped at 370 producers each year. Eligibility requires a feature with commercial theatrical release or SVOD acquisition in the past four years. Its breakfast-meeting format (i.e. matchmaker rounds with other producers across territories) is one of the best single uses of your week if you qualify.

Dress code

Is the black-tie rule real?

For the 7pm and 10pm gala screenings at the Lumière, yes. The charter requires evening wear.

But the 2025 wording is more permissive than most people assume. A little black dress, a cocktail dress, a dark-coloured pantsuit, a dressy top with black trousers are all explicitly fine. Elegant shoes or sandals, with or without a heel. No sneakers. A dark suit with bow-tie or dark-coloured tie, including a plain necktie, now complies.

Daytime screenings and non-galas only require ‘proper attire’. Business casual works fine.

What’s the 2025 nudity and voluminous-dress rule?

Two new additions to the charter, both directly prompted by 2024. After ‘naked dressing’ appeared on the red carpet (Bella Hadid in Saint Laurent at The Apprentice, Naomi Campbell in Chanel at Furiosa) and an unusually high volume of train-heavy gowns disrupted seating, the 2025 charter reads:

For decency reasons, nudity is prohibited on the Red Carpet, as well as in any other area of the Festival. Voluminous outfits, in particular those with a large train, that hinder the proper flow of traffic of guests and complicate seating in the theater are not permitted.

Enforcement was real. Halle Berry changed her Gaurav Gupta train gown at the 2025 opening, saying “I had an amazing dress by Gupta that I cannot wear tonight because it’s too big of a train.” Heidi Klum tested the boundary with a blush-pink Elie Saab train that spilled more than two metres along the red carpet, “in direct contradiction of the new Cannes dress guidelines,” and walked it anyway. Some sheer styling still appeared but with nude underlays.

The ‘decency’ clause has legal teeth. Article 222-32 of the French Criminal Code sits in the paragraph on sexual exposure and harassment, and punishes “exhibition sexuelle” in a public place with up to one year in prison and a €15,000 fine. The festival is not bluffing.

Is the flats ban over?

Officially, yes. The 2015 ‘Flatgate’ era in which women turned away from Carol and other galas for not wearing heels is over… on paper. Current wording explicitly allows ‘elegant shoes and sandals with or without a heel’.

That said, enforcement remains a bit subjective. In 2019, then-Variety editor Claudia Eller was stopped on the red carpet for wearing flats before eventually getting in, and in 2022 Dene producer Kelvin Redvers was initially blocked from the Les Amandiers premiere for wearing traditional beaded moccasins. Veterans recommend packing slim flats in a clutch and changing at the foot of the red carpet.

And no tote bags or backpacks at gala screenings - small clutches only. There are lockers near Gare Maritime which stay open until 12:30 am.

Accommodation

How expensive are hotels, really?

My past research showed, a standard Majestic room runs €815 a night during the festival versus €205 a night the week before. Carlton, Martinez and the JW Marriott all show similar 3 to 5 times inflation. The Likibu Cannes rental index spikes 177% in May, from about $221 a night year-round to around $612. Apartment rentals near the Croisette start around €190 a night, but that is a floor, not a ceiling.

I missed the booking window. Where do I stay?

In rough order of convenience: Le Suquet (old town, walkable, moderate), La Bocca (cheaper, 10 to 20 minute bus), Le Cannet (4 km north, residential), Juan-les-Pins or Antibes (20 to 30 minutes by TER train), Golfe-Juan (10 minute train), Mandelieu-la-Napoule and Mougins (car required).

Staying in Nice is technically possible but the post-10 pm screening commute back is punishing and generally discouraged.

How do I avoid rental fraud?

The Marché publishes a warning every year about scammers using Festival or Palme d’Or logos and vanishing after payment. Use the Cannes Tourism Office hotel portal, or the Marché’s official partner Immosol for apartments, which offers accreditation-linked discounts.

Getting there and getting around

Which is the best way from Nice airport to Cannes?

Five options, ranked by value:

1. Local bus 620 (formerly 200)~85 min €2.10 single

2. TER train via Nice St-Augustin~35 min~€4.60 single

3. Express bus 81 (Zou!)~45 min€19.50 single

4. Taxi~30 min€88 flat (prefectural max)

5. Helicopter charter~7 min flight from ~€1,000 per cabin (5–6 seats), or €150–€210 shared per seat

The L2 tram runs directly to both Nice airport terminals. It was inaugurated in December 2018, extended to full airport service in June 2019, and extended to Port Lympia in December 2019. BlaBlaCar rideshare from the airport typically runs €15 to €20.

Do not rent a car. Parking is tightly restricted, the Croisette is closed for red-carpet setup, and most streets are narrow and pedestrianised. A car is only useful if you’re staying in Mougins or the villages and need real flexibility.

What will a day actually cost me?

• Espresso at a side-street café is €1.80 to €3.

• The same espresso on the Majestic terrace is €8 to €12.

• Baguette at Monoprix: €1.20.

• Casual lunch: €15 to €22.

• Mid-range dinner: €35 to €60.

• Beer: €6 to €9.

• Cocktail: €16 to €22, or €25 plus at luxury hotel bars.

• Taxi across Cannes: €15 to €25.

• The €50-plus restaurants can be largely avoided by heading inland to Rue Meynadier and the Marché Forville.

Carry €50 to €100 cash for market stalls and tipping, but cards work almost everywhere, including taxis.

Parties

How do I get into the famous parties?

You don’t. The A-list parties (amfAR Gala at Hôtel du Cap-Eden-Roc, Chopard Love Night, Vanity Fair at the Cap) are genuinely invite-only and concierge-brokered, at tens of thousands of euros per seat. Don’t waste energy on them.

Realistic targets for a first-time industry attendee: country pavilion happy hours (UK, Irish, Canadian sometimes offer drinks from 5pm), sales-company cocktails at the Majestic and Martinez (Cinando contacts are your way in as a plus-one), and (the one that genuinely matters) the Petit Majestic.

The Petit Majestic is the open-air late-night bar on Rue Tony Allard, alongside the Majestic hotel. From roughly 23:00 until 4 am, it is where the real industry gossip happens, where deals get sketched on napkins, and where you need good shoes and the right energy, not an invitation. Every Cannes veteran you’ll meet will tell you this, and they will all think they are the first to tell you. They are not. But they are right.

Where do coffee meetings actually happen?

• Majestic terrace is the status default, expensive and busy.

• Martinez terrace is calmer and very expensive.

• The Grand Hotel (as it used to be known) has been rebranded as The Mondrian and so might not be as affordable and relaxed as people think.

• The JW Marriott ground-floor bar is the workhorse for catch-ups between screenings.

• Gray d’Albion lobby handles mid-tier meetings.

Rule of thumb: agree the venue when you book the meeting. The hierarchy signal matters.

The mistakes veterans told me they wish they’d avoided

When I surveyed 575 Cannes veterans in 2016, five mistakes dominated the replies. They remain the five most accurate predictions of a wasted first trip a decade later.

1. Staying the full 12 days. Attendance drops sharply after the first Sunday, when over 80% of attendees are still there. Three to five days over the opening weekend is what works for most industry first-timers, not the whole festival.

2. Overbooking screenings at the expense of meetings. You can always catch the films later. If they’re good, they’ll find distribution. Pick one or two screenings a day and keep daylight free for meetings and pavilion events.

3. Hunting buyers on the Marché floor. Buyers live in hotel offices and balcony suites. Use Cinando to book them, not the floor.

4. Wrong shoes. Cobblestones, ten-plus kilometres of walking a day, plus formal evening footwear. Veterans carry two pairs.

5. Ignoring the no-show penalty. Two no-shows is a 48-hour ban. Cancel by an hour before.

Secondary misfires include:

• Not learning the minimal French greetings (Bonjour before every transaction is non-negotiable)

• Assuming Uber availability during peak hours (it collapses around premiere times)

• Skipping meals because screenings overlap (carry snacks)

• Wearing a white shirt that won’t survive the red-carpet photographers’ lights

• Sending half-finished pitch decks in the follow-up emails from the train.

Notes

The survey of 575 Cannes veterans was conducted by me in 2016. Some responses have aged; the core patterns have not. Pricing and policy details are drawn from Festival-Cannes.com, the Marché du Film‘s official rate card, Screen Daily reporting, Filmmaker Magazine’s independent-producer strategy guide, and the Hollywood Reporter, Variety, WWD, Euronews, and Deadline for dress-code and distribution reporting.

Some prices and policies move year to year. Where 2026 figures are confirmed (accreditation opening and closing dates, the late-fee regime), I’ve used them. Where a figure is a 2024 or 2025 number and hasn’t been updated for 2026, I’ve flagged that in the text. Hotel and apartment rates are approximations and shift week-to-week; the 3 to 5 times inflation multiplier is the more reliable figure.

A caveat I owe to attendees outside Europe: most of my practitioner sources are UK- or US-based. The pavilion and Producers Network routes for other regions work differently, and readers based in Africa, Asia or Latin America should go via their national film agency or the Marché’s regional liaison for the most accurate route in.

• Top Gun 3 with Tom Cruise at 63

• Highlander with Russell Crowe at 62 and Dave Bautista at 57

• The Beekeeper 2 with Jason Statham at 58

• Jackass: Best and Last, promoted with a tearful goodbye from a cast mostly north of 50.

To see what was actually happening beneath the headlines, I took every action-tagged film from 1980 to 2025 with reported box-office or budget (4,053 leads accross 2,171 films) and worked out the age at release for every actor.

The results tell us three things:

1. The over-50 action cohort is at a historic high.

2. It is broader than any previous era.

3. And, on the evidence of the last six years, it may already be narrowing again.

The scale of the shift

More than a third of action lead roles in the 2020s are going to actors who are at least 50. That share has more than doubled since the turn of the century.

The 1990s, interestingly, is the youngest decade of action cinema in the dataset. The 1980s had a handful of durable older leads, including Roger Moore, who finished the Bond run at 58, Sean Connery, still working, and Clint Eastwood, who pivoted to Firefox at 52.

By the 1990s, most of that cohort had aged out, and the newer generation (Bruce Willis, Harrison Ford, Nicolas Cage, Sylvester Stallone, all still in their forties) was genuinely younger on screen than what came before or after.

One claim often made is that just four names (Tom Cruise, Liam Neeson, Keanu Reeves and Jason Statham) are stacking the deck. However, taking those four out of the 2020s drops the over-50 share from 37% to 35%. The cohort, in other words, does not depend on a handful of famous names.

How broad is the cohort?

One way to see the breadth directly is to ask how many actors you need to reach 50% of the decade’s over-50 action-gross.

In the 1980s the answer was two. Roger Moore alone was 30% of the decade’s over-50 action box office, and one other actor got you to half the total. By the 2010s, the answer was eleven. The 2020s, so far, is seven.

The names behind those numbers tell a parallel story.

The 1980s cohort was Roger Moore and Denholm Elliott (the latter thanks to Raiders of the Lost Ark).

The 2000s were led by Geoffrey Rush from the Pirates of the Caribbean films and Ian McKellen from The Lord of the Rings (perhaps they are “leads in action movies” but few would call them “action leads”).

The 2010s top ten was Marvel-adjacent, with Mark Ruffalo with the two Avengers: Infinity War and Avengers: Endgame entries, Samuel L. Jackson across the Avengers: Infinity War run, Mark Hamill returning for The Last Jedi, along with Tom Cruise, Liam Neeson and Sylvester Stallone.

A second way to see the breadth is to subdivide the action tag by hybrid subgenre. The over-50 share is not evenly distributed across the family.

Action-Western (64% over-50 in the 2020s), Action-Crime (47%) and Action-Thriller (43%) have aged the fastest.

Action-Superhero is the youngest subgenre across every decade. The category most associated in the popular mind with the over-50 boom turns out, in the data, to be the part of action cinema where lead actors are most reliably in their thirties.

Two forces that actually built the 2010s

Pulling the rise apart, two mechanisms show up clearly in the numbers:

1. The late-bloomer wave

2. The franchise-retention effect

1. The late-bloomer wave

This includes actors whose first lead action role came at age 50 or older, such as Liam Neeson did with Taken at 54.

92 men made their first action-led appearance after 50 during the 2010s. Some were already famous actors pivoting into a new genre, such as Bob Odenkirk with Nobody at 58, Tom Hanks with Captain Phillips at 57, and Ben Kingsley with a parallel action career in his late sixties.

Others were international leads entering Hollywood distribution for the first time, from Rajinikanth to Anil Kapoor. The cumulative effect was to widen the supply of over-50 action leads far beyond the long-tenured Hollywood names.

The 2010s also produced the first meaningful cohort of women (17) making a first action-led appearance at 50+. Before that decade, the number was tiny.

2. The franchise-retention effect

The second force is the opposite of late-bloomer debuts. It is actors who started an action franchise young and are still in it over 50.

20 actor-franchise combinations in the dataset fit that description. The longest spans, by years between first and latest entry

• Harrison Ford, Indiana Jones - 38.9 to 81.0, five films, 41-year span

• Mark Hamill, Star Wars - 31.7 to 68.2, three films, 36-year span

• Arnold Schwarzenegger, Terminator - 37.2 to 72.3, four films, 35-year span

• Tom Cruise, Top Gun - 23.9 to 59.9, two films, 34-year span

• Tom Cruise, Mission: Impossible - 33.9 to 61.0, seven films, $4.14bn global

• Vin Diesel, Fast & Furious - 33.9 to 55.8, eight films, $6.12bn global

• Hugh Jackman, X-Men - 31.8 to 55.8, eight films, $4.65bn global

• Bruce Willis, Die Hard - 33.3 to 57.9, five films, 24-year span

• Keanu Reeves, The Matrix - 34.6 to 57.3, four films, 22-year span

These two forces (i.e. more actors starting late, more actors staying on) combined to make the 2010s the broadest over-50 action cohort in the data. Marvel ensembles expanded it further, even though the superhero subgenre itself skews young.

Why do I think the 2020s are the peak?

The breadth that defined the 2010s is already giving way. The top five over-50 actors’ share of the decade’s gross has risen from 31% in the 2010s to 41% so far in the 2020s. Eleven actors reached 50% of the 2010s over-50 gross; in the 2020s, that number is seven. The trend is narrower.

Several specific 2010s mechanisms are visibly weaker in the 2020s.

• The Marvel ensemble cast members who anchored the Avengers: Endgame run have fragmented or exited.

• The John Wick franchise is in a gap year with Ballerina‘s softer reception shifting the baton to Donnie Yen.

• Liam Neeson announced in August 2025 that he is stepping back from action at 73 (while, as he conceded, still having three action films in the pipeline).

• Bruce Willis has retired.

And the late-bloomer wave itself appears to be slowing, with the pace of first-action leads at 50-plus looks, based on partial 2020s data, below the 2010s rate.

Cruise alone accounts for 13% of the 2020s over-50 action gross, with Vin Diesel, Hugh Jackman, Antonio Banderas and Dave Bautista filling out the top five. The cohort is still enormous by historical standards, but it is narrower than that of the previous decade.

It is still early in the decade, and a single breakout late debut can shift the maths. Ma Dong-seok in Korea and Shah Rukh Khan in India are arriving in the dataset in ways they weren’t before, and Bob Odenkirk may yet produce a third Nobody sequel.

But the trajectory, based on the data to date, is downward in breadth even as the ageing itself continues.

Which generation takes over?

The 2020s over-50 cohort will be in their mid-sixties by the end of the decade. Cruise is 63, Jackman 57, Statham 58, Diesel 58. The pipeline question is real, even if the data does not support the strongest version of it.

The 1970s-born male cohort (as in, Cruise’s actual peers, plus Dwayne Johnson and the Chris Hemsworth generation) is the peak for men in the dataset. 60 men born in the 1970s have three or more action lead roles by age 45. The 1980s-born cohort has produced 45.

For women, each successive decade of birth is the highest. Five women born in the 1960s reach three action leads by 45; 21 for the 1970s; 27 for the 1980s. That is the female pipeline quietly growing, mostly under 50.

Among the 30- to 45-year-old names with an anchored action credit in the last two years

• Glen Powell (36) — The Running Man (November 2025) is his first proper action-headliner swing

• Michael B. Jordan (39) — Thomas Crown Affair directing and starring, unclear franchise ambitions

• Chris Hemsworth (42) — Extraction 3 in development, Marvel commitments winding down

• Dev Patel (35) — Monkey Man and follow-ups

• Jake Gyllenhaal (45) — Road House (2024) went streaming-only

• Ryan Gosling (45) — The Gray Man and The Fall Guy both underperformed

Although these are some heavy hitters, none have yet had what Cruise had in his thirties. Namely, five or six consecutive $200m-plus hits, building audience confidence in the actor as a franchise anchor.

It takes sustained output and sustained studio backing to produce that kind of brand. The open question is whether the industry has the appetite for a twenty-year bet on a single face, as it did in the 1990s.

Epilouge

Before you run to Polymarket to bet on the ‘End of Older Action Stars’, consider that the 2020s still have space to change course. Cruise, Statham, Diesel and their peers are still working and still selling tickets. Top Gun 3, The Beekeeper 2 and Mission: Impossible — The Final Reckoning are not under-performers. Older Bollywood leads (i.e. Shah Rukh Khan with Pathaan and Jawan at 58, Rajinikanth) are filling in from outside Hollywood. F1 at Pitt’s $30m Apple payday showed that an over-60 can still launch a new property.

But the breadth that made the 2010s feel unprecedented is narrowing. The ensembles that widened the cast have fragmented. The late-bloomer wave is slower than it was. The top of the cohort is concentrating again, with seven actors now reaching half of the decade’s gross, up from eleven in the 2010s. The generation coming up behind looks talented but not yet proven at the franchise scale.

Shah Rukh Khan, at 60, has been calling himself ‘the last of the superstars’. The data from Hollywood suggests he is slightly premature, but the bench behind him is thinner than the slate would suggest.

The over-50 action-star boom is real, but on current evidence, it is beginning to tip, with the cohort ageing in place and the ranks behind it thinner than they were a decade ago.

Notes

Today’s research examined every action-tagged film released between 1980 and 2025 with a reported budget or worldwide box-office figure. That gave me 2,171 films and 4,053 lead-role observations across 1,764 unique actors.

A ‘lead’ is anyone billed in the top three positions and credited as ‘Leading’ or ‘Lead Ensemble Member’ in the cast list, i.e. broadly, the actors whose faces end up on the poster.

I subdivided the IMDb ‘Action’ tag into twelve hybrid types (Action-Crime, Action-Thriller, Action-Superhero, Action-Sci-Fi, Action-Fantasy, Action-Comedy, Action-Drama, Action-Horror, Action-War, Action-Western, Action-Adventure and Pure Action). It is a slightly clumsy taxonomy, but it lets the genre-specific patterns surface.

Ages were calculated from each actor’s date of birth and the film’s release date. About 1,800 of the 2,200 lead actors matched cleanly; the rest are excluded from the age analysis.

All box office figures are in US dollars, not adjusted for inflation. I treat gender as binary because that is how the film industry has captured the data until very recently.

Data came from Opus Data, IMDb, OMDb, Wikidata, and trade reporting from Variety, Deadline, and The Hollywood Reporter for the 2026 CinemaCon slate.