In analyzing large but quickly expanding markets like ETFs, there's an inevitable tradeoff to be made - do you want more data for fewer assets or more assets with less data?  Though I've compiled 3-year data sets and recently posted the actual data for my 1-year ETF data set for 220 ETFs, I still find it most interesting to look at the big picture right now.  That's why I recently compiled a 3-month data set with 418 total ETFs represented, about 80% of the total market right now.  It's pretty easy to argue that three months is not enough to get the big picture about anything too specific, but there are still plenty of things to look at overall and intra-specialization.

    Over the las three months, the average ETF yielded 0.07%.  The return at average plus one standard deviation is about 1.13% a day, and at average minus one standard deviation, about -1%.  Granted, it's relatively easy to see that the market is giving relatively log-normal returns, but there is some action in the higher moments driven by overall market trend and distribution of ETF specializations.