Correlation Station: The Strangle Debate

With the market high (before yesterday, at least) and interest rates very low out to forever, there's ample reason to seek out other avenues to get a decent return. Preferably one that's non-correlated to equities.

So, in that vein, here's an interesting idea: sell volatility! This, from DailyAlts:

"A 'strangle' is a multi-option strategy that involves buying (or selling) an equal number of calls and puts on the same underlying security, with the same expiration. The idea of a 'long strangle' is to capture the gains produced from volatility when you're confident volatility will occur but you're uncertain as to its direction.

This, however, is not the strategy that [RiverPark Structural Alpha Fund co-portfolio managers Jeremy] Berman and [Justin] Frankel advocate. Since implied volatility is almost always higher than actual volatility, the sellers of volatility have a systematic advantage. Thus, Berman and Frankel recommend a short strangle strategy that involves short-selling calls and puts on the same security and with the same expiration date.

In a hypothetical case study, Mr. Berman and Mr. Frankel compared the returns of the S&P 500 to those of a Systematic Volatility Selling strategy from 1990 through 2013. During that time, the S&P 500 returned a compound annual 9.9%, while the Systematic Volatility Selling strategy returned an annualized 15.2% – a difference of 53%!"

I would agree with their take that implied volatility (IV) overprices the actual risk involved. As per their numbers, IV was higher than realized volatility (RV) 86.9% of the time. I am not sure if that is concurrent or offset. By that I mean, "Did they compare IV on a given day to some measure of RV on that same day?" A better method is to offset (push the RV reading forward in time) such that the two end up covering the exact same time frame. That way you can tell whether IV ended up high or low.

Either way, though, 86.9% is high. Volatility is priced correctly if the underlying moves within the implied range 68% of the time (1 standard deviation), so it does suggest options are too high no matter which methodology they used.

They also find a mean distance of 4.5% between implied and realized, which sounds right. I default to 4 points premium, but that's just an assumption I worked with, not a hard-and-fast number. And then there's the compounded-returns edge: 15.2% to 9.9% over 23 years. It all sounds great.

I absolutely agree with their conclusion that there's a net-positive expectancy to an options selling strategy. You will have sporadic ugly draw-downs, but by and large there's an edge. You are the insurance company and you get more than enough premium the lion's share of the time to offset some accidents. I just disagree that it makes sense as a non-correlation play.

It's not correlated until it's very correlated. Namely, the short volatility portfolio is going to get hit the same time as the long stock portfolio, and probably in greater magnitude. To me, that's the very time I want non-correlation. In a way, the exact opposite works better. A systemic long volatility strategy will struggle over time, but will compound returns in a bubble up (which we don't really need) while doing well in a market implosion (which is the non-correlation we want).

Disclaimer: Mr. Warner's opinions expressed above do not necessarily represent the views of Schaeffer's Investment Research.