Sentiment Data for Hollywood
Every Monday you find it in the news, how the culmintion of thousands of hours of work and millions of dollars in production and marketing costs by film studios result in the most anticipated numbers of all, the opening weekend box office. These numbers can make or break careers, as they help to forecast whether the risks taken by the producers, production team, and others will result in success.
While these opening weekend numbers are crucial to predicting the money made by a film during its theatrical run, it turns out these numbers are only part of the story. The fact is that studios make a lot of money from films after they are at your local cinemaplex. Video on demand, Blu-ray, and even DVD sales provide a lot of revenue to studios, who must make fast decisions about which films to focus on in terms of post-theatrical production (of physical copies) and marketing.
As it turns out, opening box-office numbers aren’t well correlated with how films do months and years later when you can buy them on iTunes or at your local big-box retailer. Sometimes films that have a strong opening don’t sell that well at your local Walmart or on Amazon on demand. Other times films that have a modest box office run, like The Big Lebowski or Office Space, become cult classics that provide valuable revenue to studios well-beyond their expected shelf-life.
To see if we could improve post-theatrical forecasting, two of our economists, Steve Lehrer and Tian Xie, combined data from post-theatrical sales of films with information on sentiment and activity about these films during their opening weekend on Twitter. Their paper, recently accepted to one of the top economics journals in the world, showed that it was possible to significantly improve forecasting by X% over models using more traditional data.
This kind of work, which applies rigorous methods to sentiment data in order to address a real-world empirical challenge faced by businesses, reflects our core approach and beliefs.
For more information on our product offerings, click here
For academics who have their own ideas about how to use our data, click here