I have a basket of time series (stock prices). I want to find the N (fixed or not) time series that will best replicate the basket in the sense that combination of them will be best cointegrated with the basket.
Beside using the N series that have the best cointegration scores (ADF test) and regressing these variables against the basket. Do you know of other/better methodologies for doing that?
$\begingroup$
$\endgroup$
