Sorry, here's a bit more information...

My understanding is that two time series may appear to have a weak correlation with one another but in fact could show similar associations if one took into consideration the time lag. For instance, one time series may simply have a phase that is pushed forward a bit more than the other (but not enough to result in negative correlation). Cross correlation at 0 lag is simply a normal correlation. But, cross correlation at 1,2,3, etc. lag provides a way of looking at the associations at a particular lag. Some researchers have been interested in identifying maximal lag. I'm interested in identifying the maximal correlation given a smallish number of potential lags (maybe 1-5?). I've seen this done on occasion. Here's one potentially relevant publication. [www.ncbi.nlm.nih.gov]. Please also see question #3 and the answer here: [r.789695.n4.nabble.com].

Jatin