I have spent a great deal of time backtesting VIX Options. My object has been to seek some way of hedging a short volatility position with OTM VIX calls. Here are the details of the CBOE procedure which holds 30 day VIX calls.
What I did was to assume the sale and purchase of a single 30 Day VIX call option at the mid price each month in accordance with the same schedule the CBOE kindly produced. I used the close not the open for spot VIX. I assumed (instead of applying the above weightings to an index) that when spot VIX was <= 15 at the roll date a new option would NOT be purchased. Nor when spot VIX was above 50 at the roll date.
Otherwise, I assumed purchase of a full option if spot VIX was between 15 and 30 at the roll date and half an option if spot VIX was between 30 and 50. As per the above chart. It sounds a bit convoluted but I wanted to see what happened – after all its fair enough to map a full position, half a position or no position as per the above.
Here is what the cumulative P&L looked like. If you look at the performance of the VIX Tail Hedge Index itself, it did its job in the 2008/9 crisis (max DD 43% as opposed to around 57%) but as expected has been lagging the S&P 500 since. Not dramatically, it is true: CAGR of around 12% since end March 2009 as opposed to 14.45% for the index. Nonetheless it can be seen what an influence even such a tiny allocation to VIX calls can have when equity markets are rising and the VIX stable or falling. And I am none too clear whether those bands (15/30/50) were obtained with the benefit of hindsight. The index only began in 2011 so the temptation must have been to choose good parameters even if not deliberately. How will these parameters perform in the inevitable crashes to come?
It does not look like the VIX is predictive in the way suggested. I think my testing has been pretty accurate and I admit my method has been somewhat eccentric. But I find myself underwhelmed. And of course a 1% allocation to a portfolio consisting of a short VIX position would not touch the sides. Nonetheless, I will now take the index produced and test out various allocations as against a short VIX position. By contrast of course, far better to add an uncorrelated asset such as the Long Bond – provided the correlation does its stuff in a crisis. Long Bond/S&P correlation has varied between +.9 and -.9 over the very long term and there is no guarantee it will behave as you want it to when you need it.
At least a long bond position will increase in value over the long term, even with rising interest rates.