On a related note, Buffet vs. Protege Partners: http://longbets.org/362/
Past performance does not necessarily indicate future performance, and it should not be the only predictor, but to dismiss a fund's history as inconsequential to its future is silly.
A fund's prior performance is a useful signal that can be rationally incorporated into a risk versus reward decision process.
No, it isn't and all the funds explicitly warn you that you shouldn't do it.
The point is, funds perfomance is just a bad case of "survivor bias". Many funds are created, most of them tank and you never hear about them again, but few get lucky, make some remarkable return and get their 5 minutes of fame.
There is also one more problem with the "successfull" funds:
let's say that some fund manager actually has a secret strategy that works. At the begining, he or she just takes some initial money from investors and invest it in whatever the secret strategy suggests.
Unfortunately as soon as fund becomes popular, and people start putting more and more money in it the strategy gets thrown through the window. Why? because if you put your money in a fund that fund MUST use your money to buy stocks. Even if manager thinks that this is a bad time for buying.
Therefore, as soon as fund becomes popular it stops being strategy-based and becomes "bubble based" :)
I'm not dismissing it unconditionally, but I'm not going to believe it unless I see the data. What's the probability of a fund outperforming given it has outperformed the year before?
Yes, it's easily verifiable. It's also completely meaningless without a rational way to predict which ones will be outperforming in the future.