Bloomberg Markets magazine has an interesting piece in the current issue on a computer model used by JPMorgan Securities entertainment group to decide which movies to finance and which ones to pass on. While the accepted wisdom in movie production is that “no-one knows anything”, as William Goldman once said, JPM (implicitly) says differently.
JPM argues that too many people in the industry rely on hoary rules and dated aphorisms in making go/no-go decisions about flicks — and there is a better way, one that (shades of Moneyball) combines and integrates all the picayune data out there:
[JP Morgan] measures a film’s potential for success using a database of thousands of movies and a computer program that allows bankers to compare each new film against the financial performance of 200 or more similar movies that have gone before it. JPMorgan’s template of what a successful movie should include weighs such components as the genre, cast, director, budget and target audience against the historical performance of similar films, Miller says. “The movie business on a single picture is volatile,” he says. “Odds are that if you take a slate of 12 to 15 films, it’s almost statistically predictable. It’s quite amazing how predictable it is. You can get it within percentage points of accuracy.”