EWMA correlation calculations
Posted: September 9th, 2003, 8:53 pm
In estimating the 60 day EWMA of the correlations of pairs of large cap US stocks, what value lambda is suggested? How about if one shortened the period to 20 days? To date all that I've come up with is "'keep it between .80 and .97"...All of the literature I've read refers to lambda only in the context of persistence and reaction of volatility, not correlation... (Perhaps I needn't go no further?)Thanks in advance for your lambda and / or literature starting point suggestions. I appreciate your take on this bit of minutiae.