Like many folks, including , I would tend to agree with the following statements
- Past performance is not a very good indicator of likely future performance
- One should generally eschew managed funds in favor of low-cost index funds
However, I think a lot of folks who believe these same things are applying confirmation bias when looking at the data in a recent . Morningstar analyzes mutual funds and rates them based on their past 1/3/5/10-years performance in relation to other funds of the same type. Funds in the upper quintile of past performance get 5 stars, the next quintile gets 4 stars, etc.
Morningstar is sort of coy about whether the ratings are supposed to have predictive value. They will say that of course they only measure past performance, but there would be no way to sell these ratings to folks for millions of dollars (as they do) without there being some implication the ratings were at least partially indicative of future performance.
So the WSJ did something interesting -- they went back 10 years and took all the 5 star funds and looked at how they have done since (as measured by Morningstar itself with its star ratings). So how many 5-star funds ten years ago actually had 5-star performance over the subsequent years, and so on. And it turns out that a lot of the 5-star funds have not performed very well. This is a good reminder to us all.
BUT. Look at their own data:
Yes, the 5-star funds from 10 years ago only average 3 today. Everything regresses towards the mean, as we random walk folks might expect.
But the 5-star funds did better than the 4, which did better than the 3, which did better than the 2, which did better than the 1. This actually understates the difference, because many of the lowest performing funds in the lower star categories closed in this 10 year period, so are not in the final metrics, which likely raises the scores of some of the lower buckets because they dropped out (59% of the 1-star funds closed or merged in this period while only 22% of the 5-star funds did so).
This is actually -- to someone who doesn't really buy into the whole stock-picking thing -- a pretty impressive achievement. I challenge you to take stocks or bonds or mutual funds of roughly the same type and divide them into 5 buckets, rank the buckets by expected performance, and actually have this ranking hold for 10 years.