
Hi all, maybe someone more advanced in statistics can point me in the right direction on this.
I have a universe of stocks, a signal for each stock, and an autocorrelation at k different lags of that signal for each stock. It is very important for me to have some aggregate measure of autocorrelation of the signal as a whole, with good error bars around the estimates. The underlying time series vary in length of course, and the signal is more noisy in some stocks than others.
What is the most correct way to get an "average" autocorrelation for my signal at k different lags? Link to a book or paper would be much appreciated.
The 80/20 approach of course is averaging the autocorrelation at each lag with some standard or bootstrapped error bars.
Edit: Taking random subsamples of stocks is producing very similar AC structure, so maybe I don't need to do a lot of math here. Still curious though. 


