Date:         Mon, 12 Jul 2004 16:12:12 -0700
Reply-To:     Marky <marky@NEOSTRADA.PL>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Marky <marky@NEOSTRADA.PL>
Organization: http://groups.google.com
Subject:      Re: NEED HELP in double summing of cross multiplication..PLS!!!
Content-Type: text/plain; charset=ISO-8859-1

Hi Puddin'

I want to do exactly what you mentioned - I want to compute daily
volatility from intradaily data(tick data). So that to obtain good
estimates of realized volatility ('good' means that a mean of
cross-muliplications has to be as small as possible) I need to choose
best (equidistant) time interval and criteria is those
cross-muliplications have to be small.

Hope I made it more clear.
Cheers, Marek


pudding_man@MAIL.COM (pudding man) wrote in message news:<20040712010311.784D979002E@ws1-14.us4.outblaze.com>...
> 'Scuse me for barging in ...
>
> Like Howard, I didn't fully grasp your problem (still
> don't). You have intra-day returns on traded equity
> securities? Are you trying to derive daily returns
> from the intra-day returns? It may be beneficial to
> further explain what you're trying to do.
>
> As to the mechanics of what you ask, it is less
> problematical.
>
> Suppose that a trading day consisted of 5 hours,
> a measurement being made at the end of each hour.
> The following tested code may suggest an approach:
>
> data ee;
>  do day = 1 to 2;
>   do hour = 1 to 5;
>    x + 1;
>    output;
>    end;
>   end;
>  run;
>
> data ff(keep = day hour xday);
>  array xx(999);
>
>  do i = 1 by 1 until (last.day);
>   set ee; by day;
>   xx(i) = x;
>   end;
>
>  do j = 1 to i - 1;
>   do k = j + 1 to i;
>    xday = sum(xday, xx(j) * xx(k));
>    end;
>   end;
>
>  put day= hour= xday=;
>  run;
>
>
> Hope it hep's ...
>
>   Cheers,
>   Puddin'
>
> *******************************************************
> ***** Puddin' Man **** Pudding_Man-at-mail.com ********
> *******************************************************;
>
> ----- Original Message -----
> From: Marky <marky@NEOSTRADA.PL>
> Date:         Sun, 11 Jul 2004 14:56:59 -0700
> To: SAS-L@LISTSERV.UGA.EDU
> Subject:      Re: NEED HELP in double summing of cross multiplication..PLS!!!
>
> Ok- here is example let's assume that 1 day is divided into 5
> observations,so daily cross-multiplications are:
>
> Observations:      Output :
>      x
>      1             1*2 + 1*3 + 1*4 + 1*5 +
>      2             2*3 + 2*4 + 2*5 +
>      3             3*4 + 3*5 +
>      4             4*5 =
>      5                 = 85
>      6             6*7 + 6*8 + 6*9 + 6*10 +
>      7             7*8 + 7*9 + 7*10 +
>      8             8*9 + 8*10 +
>      9             9*10 =
>     10                  =635
>
> and so on.
>
>
>
> Howard_Schreier@ITA.DOC.GOV (Howard Schreier) wrote in message news:<200407110317.i6B3Hd214519@listserv.cc.uga.edu>...
> > So you have about 350 trading days (350 x 72 = 25,200).
> >
> > Why don't you post a mini example (perhaps 2 days x 4 intervals per day)
> > and show both the input data and expected results?
> >
> > On Sat, 10 Jul 2004 11:28:26 -0700, Marky <marky@NEOSTRADA.PL> wrote:
> >
> > >Hi Fellows,
> > >
> > >I have a problem with following matter:
> > >
> > >I have about 25000 observations which are 5-minute returns of certain
> > >equity (equivalent to 1.5 year). I need to create a formula that
> > >computes a sum of cross multiplications every 72 (here 72 equals 1
> > >day)observations. Formula is as follows: (sum from i=1 to 71)(sum from
> > >j=i+1 to 72) r(i)*r(j)
> > >
> > >where r(i) is return in time i
> > >
> > >PLEASE help me with that. I only made a formula that computes sum of
> > >multiplication over all observations, but it is useless for me  ;(
> > >
> > >THANKS for help