The VIX hit its lowest levels since 1993 this week, and that captured headlines. Whats most interesting to me is the spread between realized and implied volatility. Realized shows how much the market is actually moving, whereas implied is what the market thinks volatility will be. During sudden shocks or crisis’ (BRExit) realized volatility can trade above realized. In normal markets realized is less than implied.
The chart below shows current implied (yellow line) vs realized (blue) in the SPY. I like to think that (during “normal” markets) when the spread (the gap between the lines) narrows that volatility is “cheap”. For instance December in the chart below. In April there was major hedging of the French elections, so the IV spiked way up. The underlying market however didn’t budge.
Whats interesting about the current market is that realized volatility is essentially at historic lows. It could drift a bit lower, but its basically on life support. Implied volatility has really dropped after the French election as well, almost to the level of realized. So the situation is that realized is about as low as possible, but implied is right there with it. If they are both literally on the floor there are only two outcomes:
Volatility continues to trade sideways
Volatility gets up off the floor
I am not forecasting impending doom, I am just noting that it seems to me that the risk reward of shorting volatility is not as good as it was a few months ago. So, whether you’re long XIV, short VXX or in variance swaps I’d be cautious.
You don’t get to be in a place where you can lose that type of premium unless you are smart. I don’t think this guy is just ad-hoc betting on a market collapse. I suspect he’s shorting the heck out of volatility ETP’s (UVXY, VXX, etc) or hedging long positions in XIV, SVXY etc.
Could the guy plowing unbelievable amounts of money into VIX options be the on the other side of all the retailers plowing unbelievable amounts of money into volatility ETP’s? I think the liquidity is there to offset $80mm in VIX calls…
Andrew Thrasher CMT won an award for this paper “Forecasting a Volatility Tsunami” which is about preemptively detecting major moves in volatility. The paper is here. My review/notes are below.
He calls it the “volatility dispersion method”. Essentially he looks at the “20 day standard deviation” (he arrived at 20 days through Bollinger Bands, and notes the time period could be optimized) of the VIX and VVIX and notes that the largest spikes in VIX occur “when price dispersion is extremely low”. He then runs the same excercise with VVIX, creating two “signals” to use marking low “price dispersion”. His theory is that this is a great predictor of an impending volatility spike.
Here is my problem. His data set was May 22, 2006 to June 29, 2016. Why? I don’t know much about the market around May 2006 but I know a lot about June. That was BRExit. His data would catch that volatility event, but ignore the rest of the year. Also many of the papers and data he cites are very old – he quotes a VIX options volume stat from 2011 (its a 2017 paper). He did use 10 years of data which is quite a lot, so its possible that the rest of 2016 didn’t diminish his signal.
Volatility mean reversion at record speed indicating large demand to sell vol and keep it lower.
So, I’d be very curious to see why he chose those dates. The other stat which he doesn’t mention is realized volatility. I would be curious to see if the “dispersion” measurement would work on the S&P 500 as a whole as well? Several times he notes that there is an “anecdotal belief…that a “low” reading in the [VIX] or a large decline alone are ample reason to believe that volatility will spike in the near future.” Well, whats “low”? To answer that you have to examine realized vs implied volatility levels.
My point is that while this signal could prove useful I wouldn’t take it for anything more than face value. I would also hesitate to base much off of it going forward if the Fed substantially changes its methodology (removal of Fed put) as that type of structural market shift could ruin this signal.
For example, using this signal to buy UVXY or VXX still probably gets you nowhere.