Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: Re: st: RE: RE: Bootstrap and Technical analysis


From   "Lachenbruch, Peter" <[email protected]>
To   <[email protected]>
Subject   RE: Re: st: RE: RE: Bootstrap and Technical analysis
Date   Wed, 20 Aug 2008 11:45:01 -0700

I had a conversation about which interval to use with David Banks
several years ago and he noted that either the percentile interval or
the bias-corrected percentile interval was preferable to the normal
interval which seems to make more assumptions.

Tony

Peter A. Lachenbruch
Department of Public Health
Oregon State University
Corvallis, OR 97330
Phone: 541-737-3832
FAX: 541-737-4001


-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Eva Poen
Sent: Tuesday, August 19, 2008 9:14 AM
To: Statalist
Subject: Re: Re: st: RE: RE: Bootstrap and Technical analysis

Mahmoud,

I haven't followed all of the conversation. Just a few notes:

2008/8/19 Mahmoud Abd-El-Aal <[email protected]>:
> bootstrap (location: mean=r(mean)), rep(1000): sum var2,detail
> then i want to save all the bootstrap samples, the 1000 samples in
order
> to compare each individual mean to var1 single mean

Firstly, you don't need the detail option in your bootstrap command.
You can save the bootstrap results by using the -saving- option, see
-help bootstrap-. That would be

set seed 123
bootstrap (location: mean=r(mean)), rep(1000) saving(bootstrapsample,
replace): sum var2

You could then open the file bootstrapsample.dta and investigate the
1000 replications.

> Also when running the above command , the P value  what does it
represent?

It might help you to read the manual entry on -bootstrap-, and/or some
statistics text on the bootstrap technique. This will tell you about
the different confidence intervals that you can calculate after
-bootstrap- (normal based, percentile based, bias corrected). Type
-estat bootstrap, all- after running the bootstrap to see all
confidence intervals.

The P value refers to a test that the original sample mean is equal to
zero, very much like a coefficient test after -regress-. It uses the
estimated standard error from the bootstrap instead of the sample
standard error. A low p-value rejects the hypothesis that the original
sample mean is equal to zero. However, this is based on the normal
approximation. You might find one of the other methods more
appropriate in your circumstances.

> The basic point that i am trying to reach is how many samples from the
> bootstraped ones has a bigger mean than the value of 0.0001218, any
> suggestions?

I showed you how to do this (save the bootstrap replications in a .dta
file), but I am unsure how you would sensibly interpret such a
finding.

Hope this helps,
Eva
*
*   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/

*
*   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/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index