You can judge a book by its title: study

(CBC) - A group of statisticians in Britain has concluded that a book's bestseller potential can be predicted by its title.

Atai Winkler and his assistants have produced a complex computer model that attempts to calculate a novel's likelihood of hitting the bestseller charts.

The study was commissioned by British literary website Lulu.com, which covers the world of self-publishing. It analyzes the titles of every book to have topped the hardcover fiction section of the New York Times Bestseller List from 1955 to 2004. The winning titles were compared with the titles of less successful novels by the same authors.

"When we tested our model on 700 titles published over 50 years, it correctly predicted whether a book was a bestseller or not for nearly 70 per cent of cases," says Winkler.

According to the study, these are the common attributes of high-selling novels:

- The first word in the title was a pronoun, a verb, an adjective or a greeting.

- They had metaphorical titles instead of literal ones.

- The title involved either a possessive case with a noun showing ownership (e.g. "John's watch") or contained an adjective and a noun (e.g. "pretty horses").

According to the statistical model, Sleeping Murder (1976) by Agatha Christie is the perfect title.

But the model is not accurate in all cases. Dan Brown's The Da Vinci Code and J.K. Rowling's Harry Potter books scored low.

Bob Young, CEO of Lulu's, concedes the model shouldn't be used as a steadfast rule for authors: "If you actually write a good book with a bad title it will sell more copies than a bad book with a good title."