What use is the VIX?

Once upon a time (or, around six months ago) I decided that the volatility of US equities markets could not help but increase, and, in addition to the usual hedges I should hedge against that eventuality as well. Luckily, there was a publicly traded instrument to do just that: the VXX. Happy to have bought insurance against yet another dragon lurking in the shadows, I sat back and observed the following (the chart ends two months ago, since this shows two-month trailing volatility; I bought VXX about two months into the chart, at day 45 or so):

So, I should have been pleased with my investment. But imagine my dismay when this is what I saw in my portfolio:

The two graphs look quite different, and some statistical analysis was in order. First, let us compare the graphs of trailing volatility and the VXX, and see what we see:

There does appear to be some relationship. Now, let's start our study closer to the end of 2009. We see the following:

The relationship seems to become more tenuous. Why? Well, we recall that the first three and a half months of 2009 were very dark months indeed for the market: at that time volatility was very high (since no one was in the market, except yours truly), and returns were abysmal. In fact, this thought is borne out completely:

It turns out that outside of full catastrophe mode, the correlation between VXX and trailing volatility is quite poor (17%), and between the VXX and volatility in two months to come (for all you believers in the wisdom of Mr Market) is even poorer (around 5%). What the change in the VXX is really well correlated with is, however, is the market returns. The correlation between the change in VXX and the change in S&P 500 is a whopping -70% -- below is a scatter plot to drive the point home:

What does this all mean, given that the VXX is suppsedly backed out of implied volatility of stock options? Presumably, that in good times the market grossly underestimates its own volatility? You be the judge...