Potential Vix Forecasting Methodology?

Last week, I decided to make a volatility measure that incorporated an instrument's crossing rate along with a traditional measure of volatility.  Econometricians:  please refer to Augmented Dickey Fuller test literature for a refresher.  Long story short - I created this volatility measure, and am calling it "Crossing Rate Weighted Volatility," (CRWV).   The user has the benefit of not only measuring the amplitude of the expected deviation, but the frequency of this amplitude as well.  This is a totally different story. 

The plot thickens...

I found myself zipping through the CRWV model today, and I started working with the CRWV on the VIX (CBOE S&P 500 IVol Index).  Long story short, I found that if you take a ratio of the VIX to its own CRWV (all adjusted by a scalar constant...for aesthetics, mainly), then lag the VIX by approximately 90 periods on the CRWV...there is an interesting correlation.  I haven't had time to stress test this relationship like I normally would - but I will.  I just wanted to get this out there to start generating some ideas.  Because there is this correlation that exists at a 90-period (weekly data) lag...I'm sitting on 90 weeks of potential VIX forecast data, and I'll be forward testing this out-of-sample data accordingly.  I'm going to tinker with this idea, and I'll be giving updates to this project as they come.  Note that the correlation has been growing stronger as time marches on.  Spurious?  Maybe.  It's worth checking into.  

[Please excuse the low quality of the graphic, as I'm sitting at my terminal at work, and can't give things quite the attention/TLC that I could from home.]