susana

you if given 2 assets and if is a good diversifier could calc the new std deviation- just the way back from L1 formula w^2stddev^2 + w^2stddev^2 + 2w1w2stddev1stddev2(correlation)... and see if the new std dev is lower than the old one.
i doubt the test would get much more math oriented than that. .7 or .8 on the surface to me seems like a fairly high correlation, but it's of course all relative. i don't have the test in front of me...