Data Deep Dive: How Can We Best Predict Box Office?

Ari Levin, aka arisports, has graciously written the data article this week based on his observations of different techniques people use to predict box office revenue and how they correlate to actuals.

Hey guys, my name’s Ari.  You may have seen me around the chatter as arisports’s cineplex (still trying to come up with a more clever name.)  I’ve been reading FML Nerd’s blog for the past few weeks, and I thought I wanted to give this analysis thing a shot.  Hope you enjoy!

Every week, it seems like we are bombarded with information to help us predict Box Office, but what’s the most effective way?  Let’s take a look at some different box office estimators and see how predictive they really are.  I’m going to be using the 191 movies over the course of Fantasy Movie League history that had predictions for that week. Predictions

The folks over at give us some good predictions each week.  They seemingly know more about movies and box office totals than anyone else, so their predictions should pretty accurately reflect weekend actuals right?  Let’s take a look.


As you can see, when we look at the relationship between the prediction and the actual value, there’s a very strong relationship.  The r-value is .949, where 0 is no relationship, 1 is a fully linear direct relationship, and -1 is a fully linear inverse relationship.


We’ve seen that predictions can be pretty accurate, but what about FML Bux?  Is there something the FML team knows that we (or the predictions) don’t?


The relationship here is still pretty strong (r=.857) but not as strong as the predictions.  This is because the value of FML Bux varies from week to week, which is something we’ll look into later.  The regression also fails because for any movie under $25, it will predict a negative box office.  This is a result of only including movies with a projection, which leaves out a lot of those under $25 movies.  That as well as the fact that these have to be released on Monday as opposed to Wednesday’s predictions, it definitely looks like FML$ is a worse predictor than the pros.

Theater Count

Maybe theater counts can help us?


Or not.

Previous Box Office

This brings us to our last predictor, previous box office scores.  It may end up that these are more accurate than  However, the tests for this can only be done on movies that have already had a box score, so we need to take out all new releases.  We’ll then look at how this compares to predictions in terms of actual box office.  Ready? Go!


Wow!  That’s a very strong correlation!  Also, interestingly, this regression tells us that we can expect these films to drop an average of 51.2% each week.  What do the predictions say about this?


It turns out that is really good at predicting the box office for returning movies.  The only one I see significantly away from the prediction is Minions’s 2nd week

So what did we learn?  In the end, not much new.  We did reaffirm much of our intuition.  It’s clear now that the best predictions will come from Pro Box Office, and that they’re much better at predicting returning movies than new releases.

If you have any other suggestions to look into for predictors, please let me know!  Also, if you have any questions you’d like to see analyzed, definitely ask and I’ll look into it.

Join me next time as we look into the value of FML Bux and how they change from week to week!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s