Trading Extreme Volatility Readings With Bollinger Bands

Back during those halcyon days of early and mid-2008, when all anyone wanted to talk about was vix spikes, the indispensible counter-argument from some of us in the financial blogosphere was that arbitrary absolute VIX numbers are basically meaningless, and that relative context is the thing when it comes to analyzing volatility.

Now that the low volume summer blahs are (probably) over, maybe we’ll see some genuine premium hitting the options boards for the rest of the year, rather than being concentrated only in energy and commodities.  So we wondered, would there be any edge in taking directional trades based solely on extreme volatility readings?  Or more precisely:

Does the tendency of implied volatility to revert to the mean have any consistent correlation with price movement in its underlying?

Now the key here is what we understand by “extreme volatility.”  As mentioned above, absolute nominal values aren’t very helpful.  One of the measures most often suggested is to use instances where some volatility index is a certain percentage away from a short-term moving average.  We might test that approach in a future study.  But for this system, we looked for instances in which $VIX peeked above or below its Bollinger bands.

One thing we found is that the standard Bollinger settings (20 period, 2 sigma) weren’t helpful at all.  As Bill Luby has pointed out before, indicator inputs are made to be fiddled with, not adhered to! So we optimized a bit, and ended up with a shorter time horizon - the prior 12 days - and somewhat narrower band settings - 1.1 deviations up, 1.8 deviations down. In our test, the system buys (sells short) one contract of S&P 500 futures whenever the $VIX crosses above (below) and then closes beyond its upper (lower) band, and exits when $VIX closes below (above) its 12 day moving average.

Results

Going back 5 years, the results are quite positive.  Some highlights: this system is only in the market about 50% of the time, but is active enough, averaging 3.76 trades per month.  With a profit factor of 2.44 (the ratio of gross profits to losses), and a winning percentage of 66%, there’s some solid positive expectancy here.  One of the only unpleasant aspects is the longer hold time and relatively high number of consecutive losers - only an automated system or very disciplined trader will enter that seventh trade after taking 6 consecutive hits.

We also added some money management conditions to improve the functioning of the system:

1. When $VIX crossed the 12 day moving average - the middle line of the bands - we exited any open positions.  Regardless of its effect on performance, this helps drill down on the theory actually being tested, i.e. market action following volatility extremes.  Even if holding, say, signal-to-signal offers a better profit factor (it does), for this system we are only interested in the immediate action following an extreme reading.
    
2. We also used some limited pyramid settings so that we could profit from short trends in extreme readings.  It was very common to see $VIX peak its head above that upper band several times before settling down to the average, and this pyramiding accounts for the larger number and better win ratio of long entries.

Conclusion

This backtest seems to confirm the basic intuition that both extremes of sentiment - fear and complacency - can result in consistently profitable setups.  When implied volatility moves to extremes on a fairly short-term basis, mean reversion in the volatility tracking index seems to correlate with an inverse price movement in the underlying.

This article has been provided by Condor Options.  Condor Options is an options trading newsletter service designed to help you generate consistent 10% monthly returns with just 10 minutes a week.