In fact (as many of you guessed), it's a trick.
In 1948, psychologist Bertram R. Forer did a similar experiment with his UCLA students, having them take personality tests and then offering "personalized" horoscopes based on the tests' very specific results. That makes sense: based on something like the MMPI, you'd imagine you could return a fairly accurate description of the test-taker's personality. And sure enough, when Forer asked students to rate the horoscopes' accuracy, students gave them an average of 4.26 out of five.
Only, he gave all his students exactly the same horoscope...which was exactly the same as the one you read yesterday. This horoscope draws on fairly universal self-descriptions, light anxieties, and hopes.
This experiment is a favorite of psych 101 classes, where it tends to return results very similar to Forer's original numbers.
Shameless plug: d'you like interesting tidbit from neuroscience, psychology, game theory, zoology, and illogical human behavior? Consider picking up a copy of my new book, Brain Candy. Kill a tree. Feed my children.