- The VIX is the theoretical expectation of stock market volatility in the near future set by options market participants. It depends on option traders´ demand – the amount of premium they are willing to pay. When their perception of future risk changes, influenced by unforeseen events, VIX can jump dramatically to incorporate that shift in sentiment into prices.
- At a low level VIX is like a coiled spring and we only see the hidden energy materialize when volatility suddenly explodes. The market hasn’t had a peak-to-trough drawdown exceeding 3% in nine months and the low realized volatility in the S&P translates into disproportionally large swings in VIX caused by just slightly elevated movement in equites.
- Crowdedness in short volatility strategies resulted in a short squeeze in volatility products.
- Premiums for options on volatility products went up even more up as volatility of volatility (one of option prices´ main determinant) shot up – VVIX reached historic new highs versus VIX.
Over the long term they tend to make money, because the volatility risk premium is positive due to investor´s demand for hedges.
To analyze the actual behavior of a short volatility strategy during the rollercoaster ride between 8th and 22nd of August, I concentrate on my sub-strategy of selling weekly UVXY calls as it is very vulnerable to sudden jumps in volatility.
In short, the strategy takes advantage of the strongly negative performance of leveraged VIX Short-Term Futures ETF UVXY. I sell UVXY calls with three weeks to expiration every friday, as long as the VIX futures term structure is in contango and the S&P500 in an uptrend and let them expire worthless to earn the premium – the exact strategy rules can be found in the previous description of my volatility strategies.
- UVXY Aug25’17 30 CALL, sold on August 2nd for a normal 10% premium, was covered at a -200% hard stop (the strategy´s worst case stop loss) for an actual loss of -205,92% – without a stop loss the position would have shot up to a loss of over -350% to eventually expire on August 25th with a profit of 70%. But, as UVXY could have easily risen much further, that would have ben a desperate gamble under high pressure from rapidly rising margin requirements.
- UVXY Sep01’17 32 CALL, sold August 9th for an already elevated 15% premium, was exited at the backwardation signal (within the last trading hour before the close) for a loss of -121,02%.
- UVXY calls, with three to four weeks to expiration, lose value fast at first and then more slowly (in contrast to equity options), when VIX oscillates fairly steadily, because the possibility of a sudden volatility spike remains priced in until expiration. This means that risk is kept on board as the return potential of a short position is reduced: a position that has generated a profit of 50% (which often happens over the course of a week) will generate half the profit of a new position until expiration – it can only fall to zero. It is more effective to roll each position into a new one at a profit level between 50% to 80% – I do this discretionarily depending on how fast the call drops in value, often using small volatility spikes to patiently enter a new position.
- As a basic principal, always holding your full exposure during a strategy run is most effective to earn the highest income stream, as the inevitable loss will likely hit a full position eventually. It makes no sense to have low exposure when harvesting profits and full exposure when losses materialize.
- The most profitable time for the strategy usually is after the initial contango signal, when premiums are elevated and call value drops rapidly (often by more than 50% during the course of a few days) – the majority of VIX spikes are short term. This is of course the most psychologically challenging time as a major, volatile drawdown has just occurred. It is not productive to slowly built up to full exposure over the course of three weeks, nor to wait until a friday to start with an initial position after the signal. On the contrary, it makes sense to start with a high initial position at the highest profit probability and later lower exposure. The day the term structure returns to contango, I enter a position for full exposure (or even higher) in the last hour before the close.
- Position sizing is the most essential feature of the strategy. Trading the strategy on its own, a 3% NAV maximum exposure is fine – a 6% loss would be the usual worst case and -3% to -5% the average occurrence. In an extreme event (e.g. an overnight gap) a loss of double the worst case (-12%) should be survivable even if a string of multiple losses occurs. It is important to find a level of exposure leading to volatility that can be tolerated fairly unemotionally.
- As I trade several different short volatility strategies in my portfolio at the same time, these combined strategies have caused daily portfolio value fluctuations up to -10% at times – a level of volatility I find a bit excessive. To solve this problem, I enter a 3% initial position and then scale down to 1,5% exposure (three 0,5% positions with different expirations and strike prices) when rolling positions and throughout the run of the strategy. Other short volatility strategies are entered only after the initial position has been rolled over and exposure has been reduced.
- The initial 3% NAV position on August 11th posted a profit of 77% when rolled the very next trading day.
- Rolling over to 3 positions with 0,5% exposure at different, higher prices during the next day´s large fluctuations, they all managed to avoid being stopped out during the second volatility spike by a hairs breadth and sheer luck. These positions were all rolled over between August 22nd and 25th at profits between 50% and 60%.
- The first two of these new positions could already be rolled once again on August 28th and 31st, which led to accumulated profits of 4% NAV at a very rapid pace within three weeks, while being exposed to a potential drawdown of about 3% NAV. This supports the common observation that short volatility strategies tend to make up their vicious drawdowns fairly quickly due to the elevated levels of premium being paid – provided your capital did not shrink too much, the shock was tolerable and you are in a position to capitalize on those premiums.
- With VIX below 11 a new coiled volatility spring is building up and the pace of rolling into new positions is bound to slow down as UVXY will lose value more slowly – I found it is still worthwhile to continue the strategy at such low VIX levels, as volatility can stay low for months at a time. You just never know when the next volatility spike is coming.
Is the strategy actually worth the effort?
Definitely – the component of direct short volatility strategies (I diversify across 3 approaches with different time horizons and risk exposures, using different instruments) is my biggest driver of consistent portfolio growth in a good environment like today and the past years (2014 to 2015 posted the last big drawdowns). Protective measures are designed to stop trading the strategies in bad environments like a longer correction (2015-2016) or a bear market (2008). In backtests, including very bad periods, this leads to a Sharpe ratio well above 1 (which seems to be a sustainable upper limit for very good investors over longer time periods) – in real trading that has been much higher, but only benign environments were traded until now.
Why is the strategy so profitable?
The very fact that volatility spikes make for a painful ride is the reason that short volatility strategies will likely work going forward. They have to be hard enough to trade and stick with, so that many investors cannot do it. They have to be both structurally and emotionally difficult. If it weren’t tough, everyone would do it and the superior returns arbitraged away.
Underneath the stomach churning volatility and the dangers of being an insurer of catastrophic risks, lie several different sources of return, that are very persistent and powerful in combination:
- The volatility risk premium: the expected (implied) volatility is – on average – consistently higher than the volatility that actually materializes in the market´s movements (realized volatility). It is high and likely persistent, because investor demand for catastrophe insurance is high and they systematically and willingly overpay for it.
- The roll yield, a form of carry (the return earned from simply holding an asset, assuming zero price appreciation), is positive over 75% of the time, whenever the VIX futures term structure is in contango: Estimates of the value of VIX further in the future (VIX futures) are often higher than those closer to the present day. Selling contracts further in the future and buying them back as they get closer to maturity is a profitable trade when the VIX price stays unchanged. This drives part of the price appreciation of inverse volatility products like XIV or SVXY.
- Options (on VIX futures or volatility products like VXX or UVXY) suffer from time decay. High volatility of volatility causes elevated option premiums which a seller of OTM options earns, when the value of the underlying doesn’t change or falls (for call options).
- Volatility products like VXX or UVXY lose value through volatility drag much like a leveraged ETF. If you compare VXX and its inverse counterpart XIV, shorting both at the same time would yield about 20% per year through that drag (high borrowing costs for shorting these instruments eat up that opportunity unfortunately).
- An additional advantage is that the volatility futures term structure can be used as a precise market regime indicator.
All these elements add up in the strategy described above.
Most clearly the opportunities the strategy capitalizes on are visible in the long term logarithmic chart of 2x leveraged long volatility ETN UVXY
(data that models virtual prices of UVXY back to 2004 was used).
UVXY incinerates money, but beware of the sudden spikes! A very high percentage of OTM call options on UVXY expire worthless. Basic probabilities are highly in your favor selling these options as long as your position is not overextended.