separate positive and negative returns (Granger Causality)

[Nour88]

Dear members,
Please help..
I want to test granger causality between 2 variables (CT and SP returns) in Eviews using VAR, just like this example.. I want to separate the positive and negative returns. How can I do it with Eviews? Is there a special command for this?
I would greatly appreciate it if you give me some suggestions.

[EViews Gareth]

Set the sample such that it only contains positive values prior to running the test.

[startz]

I think the OP wants both positives and negatives. Something like @recode(x>0,x,0) and @recode(x<0,x,0)

[Nour88]

Thanks for your answers
It works and i created 2 variables (positive and negative returns).
Now, should I consider them as endogenous or exogenous variables?
Thanks in advance

[EViews Glenn]

They're explanatory variables. You won't be able to use the built-in VAR, but you can estimate your spec using the system estimator.

[Nour88]

How can I write it please in Eviews?