HP Box Office Derby: Our Final "Stand" Begins NOW!

[lcbaseball22]

Originally Posted by Wimsey

I'll reboot this thread properly after this weekend. I'm curious to see another weekend of "summer box office" to see if the increased box office from the first part of the year carries over into the summer season.

Well, this weekend has come and gone and that certainly looks to be the case:

Star Trek steered the weekend as a whole to over $149 million, which was the biggest total ever for a second weekend in May and up more than 19 percent over last year

http://www.boxofficemojo.com/news/?id=2585&p=.htm


I think this bodes well for HBP. As I said before, the poor economy should (theoretically) create an increased demand for film tickets

[Klio]

Interesting. Star Trek really *was* big, all things considered.

I hardly dared to predict that, but I sort of had a hunch.

The question is still whether this is a sign of general phenomenon, or whether this is really just an indication of the quiet but persistent Star Trek hype which seems to have reached unexpected sections of the population (at least in places I can observe).

Anyway - I guess Star Trek is the first film this summer which hasn't been disappointing as a movie (Wolverine clearly has disappointed its demographic), so the fact that it actually managed to set some record (even a rather obscure one ) could be a good sign for any other movie which manages to do the same.

[lcbaseball22]

Originally Posted by Klio

Interesting. Star Trek really *was* big, all things considered.

I hardly dared to predict that, but I sort of had a hunch.

The question is still whether this is a sign of general phenomenon, or whether this is really just an indication of the quiet but persistent Star Trek hype which seems to have reached unexpected sections of the population (at least in places I can observe).

Anyway - I guess Star Trek is the first film this summer which hasn't been disappointing as a movie (Wolverine clearly has disappointed its demographic), so the fact that it actually managed to set some record (even a rather obscure one ) could be a good sign for any other movie which manages to do the same.

Well, X-Men: Wolverine (despite not satisfying many) actually raked in about 10 million more in it's opening last weekend so I think it's a sign of films in general doing well this year at the box office. This weekend record was not due to an out of the ordinary top opening.

http://www.boxofficemojo.com/alltime.../?mo=05&p=.htm

It'll be interesting to see how long these increased box office sales continue for and if it sustains till Half Blood Prince's July opening.

[Wimsey]

Sorry for the delay: but this will be a long post, and I've been doing it piecemeal over the last couple of weeks. Although it is bad precedence, I'm going to divide it into a couple of parts, then start the list with people's predictions.

The first issue is, how should one predict the box office success of Half-Blood Prince, or any other sequel, for that matter? I am going to try to keep the jargon to a minimum, and define the jargon where I do use it. If I slip in too much jargon, then I apologize: my excuse is that I live in a world filled with other heavy users of statistics, and even when I teach stats, it is to students who have strong stats backgrounds. Conversely, if I seem a bit pedantic, well, keep in mind that I want people who do not know the equation for a normal distribution off of the top of their heads to understand this, too. Hopefully, this won't read too much like a Numb3r's script....

The second issue that I'll address in the next post are X-factors: that is, those things for which we really do not have data (or sufficient data) to examine whether they actually affect sequel performance, but that are worth considering. As we'll see, the unexplained variance from even best models still amount to 2-3M ticket sales, which would be 14-22M dollars!

The third post will be my brief summary of what worked for predictions for GoF and Order and what didn't.

The final post will be my prediction and the start of the list of predictions.

Terminology

Throughout this, I will refer to Film F1 and Film F2. F2 is the sequel to F1; in a series, this means that (say) PoA is F2 to CoS's F1, but F1 to GoF's F2. (In defiance of custom, I will not use Greek letters.... )

The data are for 20 F1:F2 pairs in which film F1 sold 40+M tickets in N. America. (That would be about $290M in the 2009 box office: but note that all analyses use estimated ticket sales instead of $$$.) Ticket sales and audience reactions are from ]Box Office Mojo, which basically has become the standard for industry use. Critical reactions are from [url=http://www.rottentomatoes.com]Rotten Tomatoes[/com]. Finally, the sales data are for N. America only: I have information on how exchange rates affected the $$$ made outside of N. America, but not information about changes in ticket prices.

The variables that we we have measured (and the shorthand I will use) are:

• audience reaction to film F1 (AR1);
• critical reaction to F1 (CR1);
• audience reaction to F2 (AR2);
• critical reaction to F2 (CR2);
• ticket sales for F1 (TS1);
• ticket sales for F2 (TS2).

Throughout this, I will discuss correlation: this is the proportion of variance for two variables (e.g., critical review and ticket sales) that is covariation. Variance summarizes the typical (squared) deviation between observation and expectation; covariance is the amount of variance that is common between two variables. If both go "up" and "down" together, then the covariance typically will be high relative to the variance, and we can reject the idea that the two variables are independent.

Models are evaluated based on the information that they provide. Technically, information measures magnitudes of change in exact probabilities of observations given models, weighted by the number of data points and the number of variables that we tweak to get the model. (Information "increases" as the score goes to zero: so what it tells you is how much is unexplained.) Another way to think about it is that information increases as the variance around expectations decreases. We also will keep track of the number of parameters used in the model: as those go up, the information will go up, but often not enough to justify using them: formal proofs exist for just how much we expect information to increase by chance alone.

Finally, when I refer to "significant," then I mean that the data disagree with an idea well enough that we reject that idea. This is done based on how we expect information to increase solely as a product of increasing model complexity. (Occam's Razor applies here, too...)

Some "stats 101" stuff...

Correlation does not equal causation!.

Or, pattern is not process! That is, if B and C are correlated (share significantly more covariation than expected by chance alone), we cannot say that C causes B or vice versa. Suppose that A causally affects B and C, and B causally affects D. Obviously, both B and C will be correlated to A; however, they also will be correlated with each other, and C even will correlate with D. There are several possible B's, C's and D's here, including: audience reactions (AR1 & AR2) and critical reaction (CR1 & CR2). These would be the B's and C's above, with Sales for F2 (S2) being the "D" and the basic properties of Films F1 & F2 represent the unmeasurable (but potentially there) "A's." In a sense, what I will do here is use partial correlates (i.e., shared correlation between two variables and a third) to tease out whether there really is an "A" and which variables might be correlated due to indirect cause and which are correlated due to direct cause. (This is type of structural equation modeling called "path analysis": I am doing an information theory version thereof.)

Effect =! determination!

In probabilistic systems, outcomes are effects of multiple variables, some of which might be important than others, but none of which completely control things. Even very successful models will have a lot of "scatter" (=residuals): that is, films that do a little better or worse than you expect given the important variable(s). The goal is to test successive models until the amount of "scatter" is sufficiently trivial that we cannot justify additional explanations (even when we can think of them). One correlate of this is:....

Outliers do not disprove simple models!

Suppose that we reject the hypothesis of independence in favor of one of simple linear correlation. However, suppose also that we have a couple of obvious exceptions to the general relationship. A common conclusion is that the simple model is wrong and that we should dismiss the whole thing. This is often incorrect: what this can mean is that we have some exceptions to the general pattern. We test this first by asking whether giving each film its own unique pattern (indicating that there are no rules) provides more information than does the a formal model. If that is not the case, we can ask if there are any particular films that really do not fit the best model at all.

Patterns in deviations can let us reject "simple" models for "complex" ones.

There are two basic ways in which this can happen. One, it could be that the relationship between X and Y is curvilinear: Y changes not just as a linear function of X, but as an exponential function, too: Y=aX+bX2, for example. Alternatively, we might have two variables responsible: Y=aX + bZ, for example. In nearly all cases, the extra "parameters" will decrease the scatter. We use information theory criteria to decide if this is justified in most of these cases. Of course, the most complicated model is a unique parameter for each sequel predicting what it earned exactly. Information theory also can tell us whether simpler models can reject that.

Results

Unsurprisingly, the Sales of Film F2 (TS2) has positive correlations with critical and audience responses. Indeed, these four parameters explain more variation in performance than actually exists: meaning that these variables are (shockingly) correlated with each other: if critics like F1, then audiences tend to like F1, and if audiences like F1, then they tend to like F2. This is intuitively reasonable. This means that some of these correlations might reflect common cause rather than direct cause. Here is a summary of the variables.

1. Audience Reaction to F1 (AR1). This has by far the strongest correlation with the performance, explaining over 60% of the variation with a simple linear model: TS2 = AR1 * TS1. However, you can do still better with an exponential model that posits still better sales when AR1 > 0.8, and worse sales when AR1 < 0.8. Note that information theory says that a unique relationship between AR1 and TS2 is actually worse than either the simple linear or the exponential.
2. Audience Approval of Film F2: this explains less than half of the variation in TS2. When we examine partial correlations, there is no relationship. When TS2>TS1, then it basically is 50:50 that AR2>AR1. In other words, "word-of-mouth" does not appear to affect sequels.
3. Critical Acclaim of Film F1 (CR1): Again, we have a significant correlation, but this correlation is explained entirely by the correlation between AR1 and CR1. If TS2>TS1, then it is 50:50 as to whether CR1 was more or less than you'd expect given AR1. So, what the critics thought of the prior film is not too important.
4. Critical Acclaim of Film F2 (CR2): This actually has the weakest (albeit still significant) correlation with TS2. Again, once we remove the correlation between CR2 and AR1, then there is no remaining correlation;
5. The Exceptions: The Lucas' Last Stand and the Sequel that Wasn't Only two films deviate so much from the "snowball" model: Revenge of the Sith and Prince Caspian. RotS greatly overperformed given Clones, and PC greatly underperformed given Lion, Witch & Wardrobe. Giving these two movies their own parameters actually improves the model significantly. This is true for no other films considered here. (Had the first Batman film sold enough tickets to make the cut, then Dark Knight would have, too: but as Radcliffe is not going to kick the bucket before July, we don't need to worry about that!)

So, what is really important? Simple: what did people think of the prior film. There seem to be some "snowball" effects. High AR1 can lead to still higher TS2: this suggests that video rentals/pay-per-view/etc. actually build up an audience for F2. (This is a new phenomenon: prior to Fellowship of the Ring, only 2 films selling 40+M tickets had sequels that sold still more tickets; since then, it has happened several times, including Prisoner of Azkaban: this does not prove the shift to films being rentable so quickly, but it is pretty suspicious!) Conversely (but not affecting HBP), if AR1 is below 0.8, then there seems to be a snowball effect towards people becoming more turned off on the series as time goes by.

Conclusion: The old business adages of Name Brand Satisfaction and Customer Loyalty seem to apply here. People see sequels if they liked the prior film. If enough people like it, then post-theater showings of Film F1 can actually increase TS2. Conversely, a middling movie will become less popular afterwards. One can debate why Sith and Caspian are not normal sequels: however, the Harry Potter films fit this model extremely well, and that is what we are discussing here. However, there still is some 2M tickets in "error bars" around the expectations: and that's a bit of money. What might the "X-Factors" be?

END PT. 1

X-Factors
1. The Economy.

When the economy goes into the tank, movie ticket sales actually tend to increase rather than decrease. "Staycations," canceling cable subscriptions, etc., all have been cited as reasons for why the box office had done better than 1 year ago. Overall, ticket sales are up 13% from last year. We have very limited information abut how this affects sequels. In fact, we really have only one example: Wolverine. Unfortunately, the Box Office boost is not helping Wolvie any: it is getting exactly what one would expect given the performance of XMen 3. The "boost" seems to be concentrated not in the blockbusters, but in the films lower in the weekly top 10. The other comparisons might be to films that had pre-established fanbases. Again, the economy is not helping here: Watchmen started off great, but rapidly sank. Star Trek has gotten off to a great start, but as this represents a complete reboot (and a great one at that!), we might expect something closer to what we saw with James Bond a couple of years back.
Conclusion We have too little data to say anything, but there is no immediate evidence that the general box office boost is helping films with established fan bases. IF Wolvie starts showing an extended tail, then we might want to revise that. (Transformers will not help as that will just be getting to the tail stage when Prince comes out.)

2. Advertising.[indent]Weren't those trailers awesome? Man, I want to see Land of the Lost so badly! Oh, wait, wrong film. WB is pushing HBP, but, again, the sequel issue rears its ugly head. New Line did not even bother advertising Return of the King, save for one short trailer only three months before the film, and a very tepid TV campaign. Why? Because as the 3rd movie, they decided that people had already made up their mind, and the only thing that the ads were good for is communicating "Dec. 17th 2003." There was not much point in doing that too far in advance. Tolkien fans screamed: "nobody will know that the film is coming out!" 60M ticket sales later..... Basically, the ad for a 6th movie is good for reminding people that it is out. The trailers will not win back Columbus Casualties or people who were never interested.
Conclusion: My favorite quote about Return of the King was that they could have Liv Tyler singing Aerosmith songs while the 4 hobbits line danced, and the film would still make $300M just due to the popularity of the first two films. People think that they know what they are getting, and ads will not convince them otherwise at this point.
3. Fan reaction.

Fan bases always assert that they have a big influence on the fate of a film. There is little data to test this, but what does exist fails to provide any support for this. Examples include:

1. Star Wars "Fan Weekend". In 1999, Phantom Menace was the #1 film for its first 3 weekends. Then, Austin Powers 2 bumped it from #1. Incensed fans at groups like this howled, and organized "Fan Weekend." That is, SW fans were going to see the film as often as they could that weekend to restore PM to its rightful place over that fake limey usurper! The feedback was impressive: everyone went 3-4 times, as did all of their friends! Surely the fanbase would boost the film! PM dropped from #2 to #4 and had a bigger proportional drop than the prior AND next weekend! Oops....
2. The Extended Two Towers Trailer. In April of 2002, shortly after the Oscars, Fellowship of the Ring ran with a prolonged "trailer" for The Two Towers at the end. As Jackson & Co. said, it really was a little "gift" to the fans. Fans exulted, and projected big things: not only would Fellowship benefit from the "Oscar Bump" but the legions of fans flooding the theaters to see the film would boost it to near-opening week levels! Tolkien newsgroups expounded with people discussing the trailer, as we all went back to the theater to see it. Then Monday's results came in and.... Fellowship showed a slight drop from the prior weekend and nothing like opening-week levels. Indeed, Fellowship did only marginally better than Oscar-winner "A Beautiful Mind," which conspicuously lacked either a big fan following (well, unless math grad students are more numerous than I think!) and no special trailer for the sequel (the as-yet-written "A Not-Bad-Looking Mind").
3. Boycotts of Fellowship and SpiderMan by enraged purists! They provided Jay Leno with good fodder, but we can see what a huge effect they had on the box offices...

Conclusion: There is not even anecdotal evidence to support the idea large fanbases influence ticket sales at the level of blockbuster movies. Even within the world of book sales, we do not see it. Remember the Harmonians and their defection for Potter fandom following Prince? It did greatly reduce the number of CoS posters: but it had no effect on sales of Hallows.

4. Loss of Fan Interest now that the Books are Done.

This is the flip side of #2, and it the concern has been voiced. We have no way of measuring this, as never before has a book series of such popularity been running at the same time as the inspired film series. However, book sales and movie sales differ in magnitude. To most people, HP is just another blockbuster, like Pirates, Spidey, Shrek, Rings, etc. There really is no difference between how films with book fandoms and those without perform. Now, HP fans do have some effect. For example, if you take the movie grades from PoA given here and compare them to those at BoM, then you would think that two completely different populations graded these films. Of course, this is not entirely true: BoM includes both HP fans AND general film goers. One can use Bayesian population models to assess probabilities of relatives sizes of the two populations, where BoM fan scores for films with similar critical ratings as PoA provide the prior probability of PoA having particular audience reactions from general audiences. For example, to get both CoS and BOM with a 50:50 split, then 100% of non-HP fans would have had to score PoA positively. No films with PoA's (very good but not mind blowing) critical reviews do so well. (Actually, no films do!) As non-fan Audience Approval becomes more probable, the proportion of HP fans dragging the grades down must drop to accommodate. (This prior probability distribution actually is a truncated: some films with PoA's critical scores actually had lower BoM scores than PoA: and that would mean a negative sized HP audience! ). One finds that the highest posterior probability for the HP fan audience was under 15%, with lower bounds going to just under 10% and upper bounds going to ~18%. Coincidentally, this predicted the book sales of Prince and Hallows very well!
Conclusion: Most of the audience isn't reading the books, so this is not an issue any more than it was for long completed works such as Lord of the Rings or James Bond, or for films without "original" source material such as Pirates of the Caribbean or Shrek.

5. Competition.

Once upon a time, competition was an important factor in the box office. However, the explosion in the number of theaters and screens, as well as the very short market life of modern films, has pretty much erased this. It is possible that two blockbusters released on the same weekend would "bump" each other: but the studios avoid that. Even a week-old blockbuster has no discernible effect on ticket sales: for example, the Big 3 of May 2007 all sold the number of tickets expected purely based on the audience reactions to the F1 films for all three, despite being released over 4 weekends. Shrek3 sold pretty the tickets expected of it in its 2nd weekend without taking into account that Spidey3 was still out there and that Pirates3 was opening.
Conclusion: Transformers 2 will open 3 weeks before. As a sequel, it will drop even faster than Transformers did (and that dropped faster than Order, if you recall). By its 3rd week, it will be a geriatric nonfactor in the box office.

6. Lack of Competition.

Prince will be the last of the major blockbusters opening this summer. As such, it might feast by being the only game in town. Really, this gets back to the Economy issue: will the box office boost extend to films for which the public already has made up its mind? My guess is "no," but this is just a hunch based on the fact that the box office dynamics of sequels seem to pay no attention to other films. Wolvie is showing this so far: we'll see what Angels & Demons and Transformers do later.
Conclusion: Color me skeptical, but without data, it is tough to say. Box offices traditionally plunge after mid-July and the big blockbusters of late June/early July never pick up steam in the absence of big releases.

7. The "troubled movie" stories.

As always happens when a film is delayed, there were lots of rumors that Prince was a "troubled" film. Titanic was a "troubled" film, too. Moreover, if you already know that you like Harry Potter films, then you probably will not care (just as you will not care what the critics or other people say): you'll see it because you like Harry Potter films.
Conclusion: I suspect that the proportion of people who pay enough attention to Hollywood to hear the "troubled movie" rumors is too low to matter and that it would not affect a sequel anyway.

8. The Weather.

Yes, the weather is important. If WB could magically create cool wet weather in the northeast and in southern California for the 3rd and 4th weekends in July
Conclusion: This is not Star Trek: my expertise in biology and stats gives me no expertise in meteorology! So, who knows.......

9. The Seasons.

It appears that CoS and GoF (as well as Toy Story 2) benefited from the swollen Thanksgiving box offices. However, so few films here were TDay films that putting in a TDay parameter just does not explain enough of the data to justify it.
Conclusion: This almost certainly is real, but as the data do not justify it, we cannot say how important it is with any repeatability. Moreover, Film F1 here is Order, and that was not a TDay film. So, we do not need to even worry about it here.

END PT. 2

This is the 3rd time that we've done this. In the past, I advised people to basically use a simple equation:

(% A+B Grades at BoM) x (Sales of Prior HP film) x Inflation + (Estimate of Audience Build by Prior Film).

This time, I've even given you an estimate of the Estimate of Audience Build! (9% +- 5%). Astoundingly, the winners for predicting GoF (QueenBeruth) and Order (several, including Jammi, I think) basically did this. What did the losers do? Well, we had big losers missed big in both directions. The low bidders disputed that Prisoner or Goblet were as popular with film-goers as BoM suggested. The usual argument was "Nobody I know liked the film." Oddly, nobody I know voted for George Bush. From this, there are two possible conclusions: Bush actually did not get ~50% of the vote in either 2000 or 2004, or my friends and family do not represent the general public. I'll let you guess which one I think to be the case, both for the elections and the films.

The "optimists" usually reasoned that as this book was "everyone's favorite" in the series, that it would that much better than the prior films. However, keep in mind that PSS sold 55M tickets in N. America. After PSS, 10M people did not care to see CoS. Now, PoA and GoF have built back the audience: but by Order the franchise has clawed back to only 78% of PSS's sales. That means that 20% of the original audience has shown no willingness to return. These "Columbus Casualties" are not going to come back easily. So, think really, really hard before betting on sales of more than 45M tickets (~$328M).


Soooo.... my bet, to get the ball rolling. I've basically come in $8-10M too high for N. America on both of the last two films. Why? I did not have the data to parameterize the "snowball" and I was overly optimistic, I guess. Order had ~88% positive marks from BOM. At this rate, we can reasonably expect a return rate of 85-95%. Despite my tendency to get burned going high before, I'm going to do it again! I'm going to go with ~95% return, which would be ~41-42M tickets. BOM has not updated its ticket prices for this year, but posters there suggest going with $7.30. So, once again, I'll round up and go with ~$305M.

International predictions get trickier. These are reported in $$$, which means that the strength of the dollar is a big factor. The dollar is doing much better against the pound than it was in July 2007 ($1.5 to the pound now vs. $2 to the pound then). That means that the same pounds in sales will generate three quarters of the dollars. However, the dollar is doing only marginally better against the Euro than it was in July 2007 ($1.3 to the Euro now vs. $1.4 to the Euro then). The dollar has slipped against the Yen. And, of course, this could all change in two months AND I have no information on relative ticket prices. ugh. So, I'll just hazard a guess of $600M assuming average dollar:local currency changes and no inflation.

There is no opening weekend again, so we will not bother with that.

Poster..........N. America....International
ArryGrotter.......$330M.........$610M
FleurDeLaPointe...$320M.........$655M
lcbaseball........$320M.........$655M
Wimsey............$305M.........$600M
KlausBaudelaire...$300M.........$550M
Klio..............$300M.........$620M
mrfutterman.......$290M.........$620M
JustAnIllusion....$285M.........$610M

I will update this list every couple of days. If you post your numbers and I miss them, then owl me.

Once the movie is out, I will (as before) post an update on estimated ticket sales for Prince vs. those for the other HP movies after the same number of weeks. As this film has a Wed. release, expect this on Wednesdays.

EDIT: ye gods, this got merged into one post! Sorry about that....

[JustAnIllusion]

Hmm, I shall try. And probably fail .

N. America - 285M
International - 610M

[ArryGrotter]

First, I think the title should be change - the delay was so last year

I think I should give this a go:

North America: $330M
International: $940M (Edit: $610? )

I am so consfused why everyone so far has predicted in the 600's for international, since OotP got just under a billion....

[lcbaseball22]

Well, I actually gave my predictions over a week ago so I'm not sure why it isn't already on the list

And I disagree with the majority of Wimsey's analysis and "theories" (aren't facts, "Columbus Casualties" for example are mythical)


So...I'm going to ignore all that and the suggested equation and stick with my own theories and numbers I previously posted

Originally Posted by lcbaseball22

My predictions for Half Blood Prince

Domestic- 320 million

Worldwide- 975 million


I think there are many factors which will contribute to HBP edging out the SS/PS to become the highest grossing HP film. The first being the poor economy. Believe it or not this should actually help out films this summer because they are still a relatively inexpensive form of entertainment. Instead of taking the family to the waterpark for example, they'll go catch a movie. If I remember correctly from Economics movie tickets are what they consider "inferior goods" and when the economy is bad, the demand for inferior goods increases. Secondly, the marketing and promotion for this film has been off the wall, with what's gotta be the most trailers for any film ever! Not only that but the test screening reviews were overwhelmingly postive and it looks to be quite fantastic. Buzz around the net, personal experience, etc shows that even those not interested in Potter are intrigued and anticipation for this film is extremely high. The delay might positively affect the film afterall. Looking at the lineup of films in the weeks following HBP, there doesn't seem to be much competition at all...the major blockbusters are all being released BEFORE so I could easily see HBP reigning supreme for 3 to 4 weeks at box office (much like TDK).

There is only 1 factor I'd be concerned with, which I think could have a negative effect- PG rating. Now I know a lot of people think common sense says that would increase ticket sales cause the film is available to a broader audience but history doesn't support this. PG-13 films in general have proven to be the more successful. My best guess for why this is that the teen/20's crowd (the largest film demographic) is turned off by the more "kiddie rating" which is something I can personally understand being that I'm 21 myself. I know my friends groaned when they heard it was PG. I'm pretty sure most of us will still go see it despite the "less mature" rating, but we'll have to wait and see...

Originally Posted by ArryGrotter

I am so consfused why everyone so far has predicted in the 600's for international, since OotP got just under a billion....

If I'm not mistaken those people are defining international as "every country other than the US"

I think you are confusing that with "worldwide" For instance, my "international" would be 655 mil

[ArryGrotter]

Originally Posted by lcbaseball22

If I'm not mistaken those people are defining international as "every country other than the US"

I think you are confusing that with "worldwide" For instance, my "international" would be 655 mil

Oh, ok, I suppose that makes sense...

[KlausBaudelaire]

I'm very pessimistic about HBP box office here in Europe.
Harry Potter is an established franchise, hands down, but it looks like interest has a little bit dried up after the DH book.

So...

KlausBaudelaire:.........$300M.........$550M

[Klio]

Originally Posted by KlausBaudelaire

I'm very pessimistic about HBP box office here in Europe.

Hmmm... it's probably true for some countries where the craze never took hold quite as much as in the UK or the US.

I think in the UK it'll do well (yes, that's Europe, too).... and in the german=-speaking countries I think the fandom is also holding up better than in italy (based on what you have said about this before).

It'll be an interesting test to see how well the fandom is doing now that the books are all published.....

[KlausBaudelaire]

Klio Hmmm... it's probably true for some countries where the craze never took hold quite as much as in the UK or the US.

Exactly, I've seen just one trailer here in the cinemas some days ago, it was the one from November.
The April trailer was dubbed, but WB didn't even put it online, or released it into the cinemas, it wasn't even uploaded on youtube, so I had to

[FleurDeLaPointe]

My Personal Prediction

Domestic: 320 million
Worldwide: 975 million

[lcbaseball22]

Originally Posted by FleurDeLaPointe

My Personal Prediction

Domestic: 320 million
Worldwide: 975 million

Well, isn't that interesting. Looks like my net stalker has decided to COPY me (exactly)

[Klio]

Originally Posted by KlausBaudelaire

Exactly, I've seen just one trailer here in the cinemas some days ago, it was the one from November.
The April trailer was dubbed, but WB didn't even put it online, or released it into the cinemas, it wasn't even uploaded on youtube, so I had to

WB should really share their Italian revenue with you for the advertising you do on their behalf.

You know who to blame if your numbers are off track because Italian box office is well above expectations

(I am kidding!!)

[Wimsey]

LC

I am perplexed. I certainly will acknowledge that to know a lot about "statistics" is to not know about a lot of techniques and methodology: I've been doing this for 20 years, and I still am learning new stuff all the time. However, surely you must agree that techniques such as structural equation modeling are appropriate for this sort of situation. After all, we have multiple hypotheses (i.e., how audience and critical reactions to two films affect the box office of the second) to test that are not mutually exclusive and not necessarily independent. Can you recommend a better approach?

I also would be curious to ask what alternatives you would propose to information theory criteria. After all, the different models being compared differ in degrees of complexity, but are not special cases of each other. Obviously, that rules out log-likelihood tests, right? But, again, if you have an alternative, then I would be glad to learn of it.

Now, I can understand if you have qualms about Bayesian probability. I'm an inveterate likelihoodist myself, and Bayesianism often leaves me a bit squeemish, so if you are from that school too, then I can relate. However, isn't our criticism of Bayesianism that too often prior probabilities cannot be justified? When we have empirical justification of those prior probabilities, then is it not hypocritical to ignore them?

But, again, given your claimed skills at statistical methods, I certainly would be interested to learn how you would test the hypotheses that I outlined above in a repeatable fashion with available data. I always am open to learning new methods and approaches, after all, and reading the views of a different stats scholar always is enlightening.


As for your statement about me possibly confusing theory and fact. [staff edit] That an apple is red is a fact. That it is red because it reflects light at a particular wavelength is theory. Both are considered to be true: surely you have not committed the layperson's mistake of confusing theory and hypothesis? And, of course, neither might be true: the color blind man would state as fact that the apple is brown, and the medieval monk would state as theory that the color reflected the animus of the object. We might be wrong in what we see and todays theories might be tomorrow's alchemy. (Of course, in this case, and as I am sure I do not need to explain to you, the methods that I used explicitly take into account uncertainty in proportions: the evenly weighted coin does not, after all, flip 50% heads.)

At any rate, there certainly is no theory in what I wrote: it is purely about testing existing hypotheses, as I am sure that you recognize. All the theory is the accepted theory of cinema (i.e., that properties of the film accept how audiences and critics perceive it).


To this end, I am sure that you realize in retrospect [staff edit ] that "Columbus Casualties" are mythical, unless you are suggesting that somehow the box office totals showing that 20% of the people who saw PSS did not come back to see COS. (This likely was even lower overseas: I can only correct for the decreasing value of the dollar, and if there was any inflation, then over 20% of "international" goers did not return.) 10 million people in North America alone is hard to call "mythical": I do not recall anybody suggesting that CoS's box office declined because the Sasquatch population decided to skip it, for example.

Data are, of course, data: and there is no color blindness here. "Columbus Casualties" is the label for them dubbed by (if I remember) an analyst at BoM, after surveys in 2003 revealed that PSS viewers who did not come back to see CoS generally did not like Stone because it was too slow, too long, and too lacking in story ("point."). Those were the facts as WB acknowledged them in 2003. The only "theoretical" leap is that it is accepted theory that these are criticisms of the direction of the first film.


Oh, and sorry that I forgot your prediction: but, as I am sure that you can related, busy people have lots to do. Remembering such details a week later is asking a bit much of me!

[KlausBaudelaire]

Originally Posted by Klio

It'll be an interesting test to see how well the fandom is doing now that the books are all published.....

I'll make an example of a book ,adapted for the cinema years later its release: Veronika decides to die.

Of course this book, who came out in 1997, has never been huge as Harry Potter, but it works perfectly for what I am trying to explain.
Twelve years ago, Paulo Coelho was at his best, having published The Alchemist some years before, and with a very spot-on marketing, Veronika decides to die became soon a bestseller book, translated in almost every language.

In 2008, director Emily Young started shooting the adaptation starring Sarah Michelle Gellar, David Thewlis (Lupin!) and oscar nominee Melissa Leo.
The film promised to be very faithful to the book, exception for the setting (NY instead of Slovenia, so to please the american audiences) and Paulo Coelho himself promoted the film in the very trailer.

In case you're wondering when the movie will be released, the answer is most likely never: its release keeps on being pushed back.
A film inspired by a bestseller, with three famous actors, a talented filmmaker, and promoted by Paulo Coelho promised to be a cult...what went wrong?

Simple enough, the screening-tests had an awful response--producers thought the book's success and Coelho's name were enough to gain immediate success--but people didn't even remember a book with this title and a writer with such name, worst of all, films with similar plots, like Girl:Interrupted came to steal the show, and the Veronika film was labelled as a bad carbon-copy.
Furthermore, Sarah M. Gellar's latest movies bombed terribly at the boxoffice, and no one invested on her as protagonist.

What's this got to do with Harry Potter (apart from D.Thewlis)? simple, Veronika is a case were everyone sponsorized the movie as the adaptation from the bestseller booook!!! but the book in question was released eleven years ago, and Joe and Jane public couldn't care less.
Of course, the Veronika's troubles will never trouble HP films, for they are established and recent successes, but they will obviously lose a portion of the readers who were unhappy with the Columbus films, or unhappy with the DH book.

[lcbaseball22]

Originally Posted by Wimsey

LC

I am perplexed. I certainly will acknowledge that to know a lot about "statistics" is to not know about a lot of techniques and methodology: I've been doing this for 20 years, and I still am learning new stuff all the time. However, surely you must agree that techniques such as structural equation modeling are appropriate for this sort of situation. After all, we have multiple hypotheses (i.e., how audience and critical reactions to two films affect the box office of the second) to test that are not mutually exclusive and not necessarily independent. Can you recommend a better approach?

I also would be curious to ask what alternatives you would propose to information theory criteria. After all, the different models being compared differ in degrees of complexity, but are not special cases of each other. Obviously, that rules out log-likelihood tests, right? But, again, if you have an alternative, then I would be glad to learn of it.

Now, I can understand if you have qualms about Bayesian probability. I'm an inveterate likelihoodist myself, and Bayesianism often leaves me a bit squeemish, so if you are from that school too, then I can relate. However, isn't our criticism of Bayesianism that too often prior probabilities cannot be justified? When we have empirical justification of those prior probabilities, then is it not hypocritical to ignore them?

But, again, given your claimed skills at statistical methods, I certainly would be interested to learn how you would test the hypotheses that I outlined above in a repeatable fashion with available data. I always am open to learning new methods and approaches, after all, and reading the views of a different stats scholar always is enlightening.

First of all, I think you have got the wrong impression. I am by no means a “stats scholar” Simply an engineering student who’s taken some upper level stats courses. Enough to understand the terminology and such and draw my own conclusions. No, I don’t have any issues with your methodology…it all seems pretty sound to me. And sorry to disappoint you, but I don’t have any alternative methods or suggestions.

However, I disagree with many of the factors you have analyzed and/or the conclusions you have drawn from them. But many of these are the “X-factors” such as economy, advertising, competition, etc. I would dispute the significance of such, but as you noted there is little concrete evidence…so it’s hard to draw conclusions either way. I have a feeling though these factors will have more impact than you think.

And call it a hunch, but I don’t believe Half Blood Prince will follow any generic patterns or trends. I see the Harry Potter franchise in general as fairly unique and something which can’t really be compared to any previous examples, other than perhaps itself. I suspect this shall be even more the case with the 2-part Deathly Hallows film, as it will be breaking new ground in cinema. But back to HBP, naturally as I don't believe in any sort of established trend or agree with your suggestions, I don't feel that equation will yield a very accurate prediction.

Perhaps you’ll end up kicking my *** and I should have followed your advice and used a more traditional approach () but we shall see

Originally Posted by Wimsey

To this end, I am sure that you realize in retrospect [staff edit ] that "Columbus Casualties" are mythical, unless you are suggesting that somehow the box office totals showing that 20% of the people who saw PSS did not come back to see COS. (This likely was even lower overseas: I can only correct for the decreasing value of the dollar, and if there was any inflation, then over 20% of "international" goers did not return.) 10 million people in North America alone is hard to call "mythical": I do not recall anybody suggesting that CoS's box office declined because the Sasquatch population decided to skip it, for example.

Perhaps “mythical” wasn’t the right word to use here. Maybe “hypothetical” would be better, I don’t know…I’m not the best with words, my skills lie more in math/science than English/writing Anyways, I simply think there are way too many variables involved when it come to box office intake, to state the cause for the decline was due to poor films or whatever. Afterall, SS/PS and CoS received nearly as much praise as any of the other films, as I demonstrated in a previous post- http://www.cosforums.com/showpost.ph...&postcount=178

I'm sick and tired of some blaming Columbus and accusing him of nearly jeopardizing the film franchise before it hardly got off the ground (800 mil still ain't too shabby). And the first film just so happens to still be my favorite thusfar, so naturally I’m going to defend it.

[Wimsey]

One more bit of information. International box offices are reported in US dollars. So, 100 lbs in 2001 means a very different number of dollars than does 100 lbs in 2002 because of the ever changing value of the dollar. That actually was huge because the dollar plunged during most of this decade, although it recently has "rebounded" (or had the Pound, Euro and other major currencies drop to its level) recently.

This actually affects the apparent shifts in the box office quite a bit. The table below gives the International box offices as reported ("Int") and then adjusted ("Adj") based on a weighted average of the major currencies for the regions. The drop (∆Int and ∆Adj) shows major effects.

Film.....Int....Adj....∆Int....∆Adj
PSS....$659....$991......•.......•
CoS....$617....$837.....94%.....84%
PoA....$546....$610.....88%.....73%
GoF....$606....$712....111%....117%
OotP...$646....$646....107%.....91%

The "Columbus Casualty" pattern is much more pronounced when one accounts for the fact that the weakening dollar inflated the amount of dollars that Pounds, Euros, Yens, etc., generated: the first two films brought back much lower audiences. The extra drop for PoA likely reflects the additional fact that ticket sales in Europe and Asia are lower in summer than in the winter: unlike the US, where movie-going is associated with summer, movie-going is a winter activity elsewhere. (Two hypotheses exist for this. One is that Europeans and Asians are more apt to spend summer weather outdoors. The other is that the American tradition of summer movie-viewing evolved from outdoor "drive-in" movie viewing; this never was big in Europe or most of Asia because those countries are further north and it is much later when it finally is dark enough to show a movie outside!)

As a corollary, PoA brought back a much bigger audience in internationally than it did in N. America. However, that would have been due partially to elevated November/December ticket sales in those countries.

Regardless, it does make Goblet's performance (getting ~91% of the local currencies) pretty impressive.

Again, I have no data on inflation. Presumably, ticket sales decreased as movie tickets almost certainly increased in all of these countries over time. Nobody at BOM seems to know where to get such data, so it must be pretty obscure!

The dollar is close to 2001 levels for the pound, but the Euro and Yen both re doing ~20% better relative to the US dollar than they were in 2001. However, the Euro is about 15% worse than it was when Order was out.

Make of that what you will!

[Sacred_Memories]

I'm a member in the box office mojo forum. It's really good. I'm The Glacial Prince. Anyway, I personally see this movie doing $324M US, and while all HP movies have managed to double their domestic gross internationally, I think this one will fail to double it because of the rising dollar value.

Right now, I see $324M domestically and $605M overseas for a total of $929M.

[Wimsey]

Sacred: it might be better to state that all the other currencies have dropped rather than that the dollar has risen!

At some point in the next week, I'll post on the "half-lives" of sequels. The pattern is should make intuitive sense (and mirrors that of the books), but HP is breaking into new territory for "blockbuster' sequels in terms of the number of films still selling 40+M tickets.

Updated list:

Poster..........N. America....International
ArryGrotter.......$330M.........$610M
Sacred_Memory.....$324M.........$605M
FleurDeLaPointe...$320M.........$655M
lcbaseball........$320M.........$655M
Wimsey............$305M.........$600M
KlausBaudelaire...$300M.........$550M
Klio..............$300M.........$620M
mrfutterman.......$290M.........$620M
JustAnIllusion....$285M.........$610M