Subsequent to their algorithm change in 2002, Christine Benz in her 2005 book Morningstar's Guide to Mutual Funds, said, “…the star rating is based on how the fund did in the past. It won't predict short-term winners.” 

Finally, Don Phillips, in 2010's “Star Wars, the Sequel,” said “The Morningstar Rating for funds is a grade on past performance. Period. No one at Morningstar ever claimed that the stars have predictive power.”

But now Morningstar's tune might be evolving. 

A spokesman for Morningstar told me, “…after we changed our methodology in 2002,… we did find some moderate predictive value.”

Edwin Choi, however, in his article, “Morningstar Ratings: Useful or Useless?” analyzed a 2010 Morningstar study. He concluded the study showed the expense ratio was a better predictor of outcomes than the Morningstar Rating.

Two Pace University professors ran an independent study and reported, “…we find widespread support for the notion that the new Morningstar rating system can predict future performance…”

Yet Morningstar's own licensing agreement, after sketching out how the Star Rating is calculated, states “Past performance is no guarantee of future results.”

When asked if it believes the stars are predictive, a Morningstar spokesman said: “That depends on your interpretation of the word 'predictive' – [does it mean] guaranteed or associated with good outcomes?” Morningstar's Star Rating doesn't 'guarantee' any results. We believe the star ratings are associated with good outcomes, but we've always said we don't want people buying funds simply because they have 5 stars.

So do they or don't they? In Morningstar's eyes, the answer appears to be an unequivocal, “Yes.” 

Schrödinger would be proud.