Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


dickreuter
Posts:
56
From:
London
Registered:
10/18/08


Re: Wilcoxon test?
Posted:
Mar 11, 2009 8:57 AM


On Jan 13, 5:24 pm, Robert Dodier <robert.dod...@gmail.com> wrote: > On Jan 12, 2:09 pm, NickPick <dickreu...@yahoo.com> wrote: > > > Interesting view. But shouldn't statistical significance lead to > > practical significance in some way? I.e. when I have a trading > > strategy which gives statistical significant outperformance over an > > index, shouldn't I be able to draw practical benefits from that? > > Whether some result is practically significant can't be answered > by a detour through statistical significance, as there is no > necessary relationship between practical and statistical > significance. > > If some result is statistically signficant but not practically, > why would you care? How about the reverse? > > Suppose you are trying to explain to your boss. You say, > "The difference between method 1 and method 2 is statistically > significant! Hurray!" Your boss says, "Can we make money on this?" > What's your reply? If you know your boss wants to make money, > why would you ever bother with statistical significance again? > > FWIW > > Robert Dodier
In fact what I'm trying to determine is whether a hedge fund actually outperformed the index (ie. generated so called alpha). IMHO this can be determined in looking at each month's returns of the fund and determine whether they are statistically significanly different from its benchmark (i.e. the index) over a certain time. If it is not, the hedge fund has not outperformed and its strategy is obviously not working.



