I received a tremendous amount of value out of Mr. Cain Thaler’s thought process for PSN.TO and I hoped someone else could similarly benefit here.
A while ago (the press release states March 14, 2012), I came across an interesting new offering from the CBOE: $VVIX, or a volatility index for the $VIX. At the time, the product had no live-historical data nor had the CBOE released any back-tested history. There was this one Excel file, however; which was clearly made by an individual for his own purposes (rather than broad / professional distribution) and nowhere near as comprehensible as it is today.
There wasn’t much I could do in terms of research at that point short of re-engineering the data myself; I determined that it wasn’t worth my time and I waited for the CBOE to produce the data. This tab in my browser remained open for two, maybe three, months. At first, I checked in on it every now and then; and while the historical data eventually made its way to the site, I ultimately decided to close the tab and revisit this index at a later date.
That date was today. Last night, Woodshedder posted this link in the news section of iBankCoin. In sum, it was a write-up about a research piece based on previous research stating that “portfolios consisting of stocks with low vol-of-vol have consistently ouperformed their more uncertainty-laden peers.”
These individuals did an extensive amount of work–correcting for variables–and came to the conclusion that lower vol-of-vol outperforms higher vol-of-vol. I now know I need to immediately dedicate time to $VVIX. But before I get started, I want to set some expectations for the process (not the results).
Here’s why: before the study is completed, I will begin to understand what the likely results will be. The conclusion may or may not be very strong and I might be tempted to add some parameters / qualifications / modifications in order to strengthen the conclusion. This is a slipperly slope that may be extremely detrimental or extremely beneficial, but my expected process can help determine whether those modifications are appropriate or not.
I’ve noted the following and I will produce the research at a later point:
- Unlike the research paper, $VVIX will be comparing one portfolio to itself ($SPX) over time.
- Over time, volatility standards will change, thus the model will have to adapt to relative conditions
- Adaptation to vol-of-vol (i.e. erratic) behavior by way of moving average is likely to produce many bad signals
- Adjusting changes in $VVIX by changes in $VIX may or may not be appropriate
- Adjusting changes in $VVIX by changes in $SPX may or may not be appropriate
- Results may be one-sided (i.e. an extremely high $VVIX may indicate poor future returns, but everything else has no predictive power)
I never know where my next study will come from or what results it will produce. A previous study of mine didn’t really produce significant results on its own (and I stated as much in the title). However, that concept proved useful in a future topic, which did produce some interesting results. In this case, it took 5+ months and 4 degrees of separation before my research even got started.
Remain open to the possible connections.