This posting raises questions at several levels. Here are three.
1. The Brock et al. 1992 is a minimal reference of the kind you are
specifically requested not to make in the FAQ.
2. You are comparing two time series models. Bootstrapping of the
standard kind is predicated upon independence. Bootstrapping with time
series data isn't easy and it isn't especially well supported in Stata.
I don't have more positive suggestions on how to do the bootstrapping; I
just warn that what you are doing is problematic. On the other hand,
there seems no reason why you should not compare distributions
graphically.
3. Pleas that your case is somehow especially deserving are I think
generally considered out of order. See e.g.
<www.stata.com/statalist/archive/2008-04/msg01286.html> and the
supporting thread. Sure, you are anxious to finish your dissertation,
but most of us have been there too, so please lay off the emotional
blackmail.
Nick
[email protected]
Mahmoud Abd-El-Aal
I am trying to replicate the calculations in Brock, Lakonishok and
Lebaron
1992 paper about technical analysis.
I am using two null models, the random walk with a drift and the AR 1
in the study, they produce tables that compare the simulated means and
standard deviations from the return series of the technical rules with
the
original Index returns.
My problem is basically comparing the means and standard deviations
using
the bootstrap
If i type the command :
bootstrap (location: mean=r(mean)), rep(500): sum var1, detail
this gives me only var1 results and i have to compare var1 and var2
i tried to preform a ttest using bootstrap in order to see if the two
means are equal but this did not help as i encounter problems running
the
t-test with bootstrap
All in all i am stuck and this is the last part for me to finish my
masters dissertation, so any help is needed. thanks all.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
