Ochotona, kbg,ochotona wrote: ↑Mon Jun 17, 2019 1:48 pmNovell didn't share the data with me, but he did plug the synthetic USMV data into his proprietary model and ran in it place of SPY, and it was better. The GEM public model is also better 2013-2019. I also looked at the Invesco vs. iShares min vol ETFs, and iShares were far and away better. Yup, this is a cool twist.
In fairness to myself... I bought USMV buy-and-hold for the Mrs. before all of this 2018-2019 volatility erupted. But now that we've had the short VIX trade destroyed, and the Christmas 2018 end of the world, plus the late 2015 thing, I think we have some nice views into how the ETF performs under stress. It works as designed.
The only thing I worry about has been voiced publicly, and I think it's happening now... when things go pear-shaped, everyone piles into USMV. Everyone gets on the same side of the boat. It's way oversold now, possibly a bad deal if you buy it today.
I checked the underlying index for these ETFs; it appears that they are all based on regular minimum volatility indexes from MSCI and are not proprietary; see:
https://www.ishares.com/us/products/239 ... tility-etf
https://www.ishares.com/us/products/239 ... tility-etf
https://www.ishares.com/us/products/239 ... tility-etf
Go to any of these links, choose the "Key Facts" tab, and see what the underlying benchmark index is.
I have downloaded the appropriate indexes from the MSCI website and uploaded a zipped file of them to:
http://www.filedropper.com/msciminimumvolindexes
The US minimum volatility index goes back to mid-1988 and the ACWI minimum volatility index (which includes all the countries of the world including the USA and everywhere else) goes back to mid-1993; I also included the EAFE minimum volatility index (which underlies the ETF EFAV) and which goes back to mid-1988 as well; you can either use this to backtest a EAFE-US minimum volatility DM back to 1990 or--provided you can figure out how to weight EAFE vs US each year depending on roughly how much of the world's market cap the US was vs the EAFE--use it to create a quasi-ACWI minimum volatility index back to the late 1980s for a US-ACWI minimum volatility DM (using the EAFE + US "synthetic ACWI index" until mid-1993 and then the actual ACWI one after that); otherwise the US-ACWI minimum volatility DM will have to start in 1995 since the ACWI min vol index starts in mid-1993 and 1994 was the first full year to give you twelve months of returns to look at.
BTW, I checked (just from mid-2017 onwards to May 2019) both the monthly MSCI US min vol index returns vs USMV and the monthly MSCI ACWI min vol index returns vs ACWV; the monthly correlations on the returns were 0.9990 for the US one vs its underlying index and 0.9971 for the ACWI one vs its underlying index (in other words an almost perfect correlation in both cases); I think that shows that these ETFs are indeed based on these indexes, track them fairly well, and as such said indexes can safely be used to extend the backtest before 2013 and even before 1999.