Yeah, I trade a different asset class (rates) so I don't really believe much in any sort of vol prediction (use EW vol,but mainly to check if current implied vol is consistent), but don't use anything more complex. In rates, events (eco announcement, auctions ) seem to be effecting vol more then previous vol, MR etc. I regress the log changes in forward IV to the log changes in RV, usually 20-day changes. To make my life more miserable, I have to incorporate it into a model that also predicts the reset skew and dependence of vol on the steepness of the yield curve. What other tests do you run on term structure trades?
I quote var with the swap, mid on the VT futures (or pinned VIX straddle), cspline LR on the data and solve for a short duration vol figure. Admittedly, I don't use it much.