Does Twitter Predict Box Office Better Than Industry Analysts?

For months new, industry analysts have wondered about "Twitter Effect." Does Twitter have the capacity to change people's perceptions of films? Did it make Bruno fail at the box office? Did it alter the outcome of the Academy Awards?

Whether or not any of that is true, some recently-released research by HP (via Mashable) purports to demonstrate that Twitter is better at predicting box office than currently-accepted methods. Hit the jump for some more details

Predicting box office returns has become a cottage industry in Hollywood, to the point where investors are thinking of establishing a futures market for it. One of the best methods of predicting box office is the Hollywood Stock Exchange (HSX), which is essentially a massive game where people buy "shares" of actors, directors, and movies. With over 200,000 players, the HSX has proven to be an excellent prediction market, with some exceptions (e.g. According to HSX, Avatar was supposed to bomb). But is Twitter the superior prognosticator?

Researchers analyzed the number of tweets about a film and the rate at which they were being produced then cross-checked it with the box office returns of 24 diverse movies (the list included Transylmania, Twilight: New Moon, and Avatar). Using 3 million tweets from the week before a film was released, they were able to create a linear regression model for predicting box office returns. They found that their model was able to accurately forecast box office performance. Moreover, not only did their model outperform the HSX-model, it was actually useful for predicting future HSX prices! The researchers also analyzed whether tweet sentiment was positive or negative after a film was released, and discovered that positive sentiment was correlated with better returns.

I think the research does make a couple of spurious assumptions, specifically surrounding subjectivity. For example, in analyzing the sentiment of tweets, the authors write, "Positive sentiments following the release can be considered as recommendations by people who have seen the movie." As we all know, whether or not someone has seen a film will often not affect their capacity to talk at length about their extreme feelings about it, especially on Twitter.

Still, it's fascinating to see that Twitter might have more industry impact aside from letting you know where Ashton is spending his afternoon.

Discuss: Do you think Twitter can be used to accurately predict box office returns? What has been your experience with your own Twitter stream?