sample in a group

[martingale]

Use Eviews 7.

I have a group consisting of daily returns of individual stocks. Need a sample that shows only daily returns of individual stocks above market return and make a new group with these observations. How can this be done? A spesific solution would be greatly appreciated.

[EViews Gareth]

You'll have to be more explicit as to what you're trying to do.

[martingale]

Example
group members:
Company1 return 0,01
Company2 return 0,04
Company3 return 0,04
Market return 0,03

need to sample so that companies above market return form a new group, that is company2 and company3

[EViews Gareth]

Sorry, that still isn't clear.

You have a group. Roughly speaking, that's a matrix - each row is an observation, each column is a series. You want to set a sample (which selects observations, i.e. rows) so that the members of the group (i.e. columns) change. You can't do that.

[martingale]

just did it in Excel

[trubador]

Code: Select all

'This part generates random data to simulate the current situation as I see it. wfcreate u 100 group full !market_ret = 0.01 for !i = 1 to 10 series y!i = nrnd full.add y!i next 'This is the actual part that is supposed to solve your problem (from my understanding of course). group sample for !i = 1 to full.@count %s = full.@seriesname(!i) if @mean({%s}) > !market_ret then sample.add {%s} endif next

[martingale]

Thank you! Some adjustment is needed.

I have individual files of daily returns for companies and the daily returns of the market as a whole for the same period of time.

The name structure for the individual companies is:

R_N_companyX_P_

where the component companyX varies from company to company.

The name for the market returns is:

r_msnwayl

The group name which includes all the above is:
R

I need EViews to:
Compare each company daily return with the market return and create a new group with the observations which are greater than the daily market returns for each day. I have 1445 daily observations and 207 companies.

The purpose of this is to run a test of the cross sectional absolute deviation(CSAD):

CSAD(t)=Sum(i=1,N){(1/N)absolutevalue(R(i,t)-R(m,t)}

where R(i,t) is the return of company I at day t and R(m,t) is the market return at day t, and N is the number of observations that fulfill the requirement Ri>Rm at a given day.

The prediction of rational expectations is that in the equation:
CSAD(t)=gamma(0)+gamma(1)absolutevalue(R(m,t)+gamma(2)(R(m,t))^2
gamma(1) is positive and gamma(2) is zero.

If gamma(2) is negative and significantly different from zero the interpretation is herding behavior in the market.

[martingale]

a solution was found , you find the research done and the programming code used in e-views here: http://brage.bibsys.no/xmlui/bitstream/ ... ilidis.pdf