Edit May 2018: I have posted comments after large volatility events in August 2017 (including important strategy insights and adjustments) and February 2018, which highlight how these volatility spikes played out in practice in my portfolio.
I have also radically simplified my direct volatility trading strategy, making it super easy to run. You can find the detailed new strategy rules as well as the old (to keep them on record) below – both variations have served me well, yielding similar returns.

• Volatility always returns to its long-term mean (mean reversion effect).
• Volatility tends to spike briefly (usually when the stock market slumps), followed by lengthier downward trends.
• Volatility forms volatility clusters (regimes), the best predictor of future volatility is current volatility.
• Volatility strategies manifest a tendency to rebound very quickly after drawdowns, because of rising premiums in bear markets. That implies, that the returning calm after a severe storm is the most lucrative environment for being short volatility, but also the hardest to stomach.

• Selling out of the money put options on index futures & ETF (insurance) across asset classes and selling far out of the money call options on individual securities and ETF (lottery tickets). These carry the highest premia as investor demand for these options is highest and they systematically overpay for them. Studies show that it also works well to combine index short puts and short calls into systematic strangles, that make a portfolio largely market neutral. This type of strategy captures the volatility risk premium while avoiding to make a directional bet.
• Trading volatility directly through VIX Futures, options, ETN, ETF and their options. Volatility ETP like VXX are some of the most efficiently value-losing securities out there and money can be made taking the other side. The danger lies in the vicious spikes these securities can have to the upside.

• Trading universe: all global asset ETF and selected futures with liquid options including equities, fixed income, real estate, commodities and currencies, plus individual securities that are doing badly as underlyings to sell lottery-type calls on. Finding suitable ETF is still done by manually looking for option availability and tolerable bid-ask spreads.
• Basic entry rules: to sell puts (for short calls use reverse criteria) do a monthly check: if ETF price is above its long term moving average (150 to 300 days, I use a 275 day SMA to be a bit on the longer side from the commonly used 200 day SMA) and 12 month momentum has risen to the top 20% in the last 1-3 months (this screener works well), write puts on all ETF passing the screen, checking for diversification, liquidity and capacity of portfolio allocation.
• Adaptive portfolio allocation: Calculating with the effective exposure to the underlying, I allocate between 50% and 150% of capital (NAV) to short options (this includes direct volatility trading described below) with hedging in place to neutralize some market direction. I´m scaling down exposure in bear markets and high volatility regimes and cap the amount of option writing at my current maximum allocation. The more positive trend and momentum candidates pass the screens, the higher my overall allocation and vice versa.
• Position size, strikes and time to expiration: Each position is calculated by writing option premium for 0,5% of capital, which automatically results in scaling positions approximately by the underlying´s volatility (the underlying´s volatility determines option prices). Strikes are 2% – 5% OTM for puts and around 5% OTM for calls. Expiration dates are spread equally from 2 to 4 months to diversify across time, new positions are placed in the slots, that are opening by exiting existing positions.
• Diversification and market neutrality: ETF are selected by liquidity and diversification across sectors, countries and asset classes, avoiding concentration in any particular area. When allocation capacity is full, no additional options are written until individual positions are closed. The market´s direction is hedged by writing calls on downtrending securities, depending on the current trend (conceptually this results in a strangle spanning different, correlated securities): the more ETF show a positive trend, the more the option portfolio will be positively tilted. When more markets trend downwards short call exposure is scaled up and overall leverage scaled down.
• Exit rules: option positions are exited at a 50%+ profit or via stop loss at -100% or at the break of the long term moving average of the underlying.
• Strategy statistics: I have made a lot of changes, but since late 2015 the win rate was 75-80% with an average loss at -80% and an average win at 43%. Typical for a negative skew strategy is the large amount of small wins with few, larger losses, but because of the benign market environment and few major adverse events these statistics are not representative of a full cycle.

• VIX Futures contracts began trading in 2004
• VIX options began trading in 2006
• VIX Short-Term Futures ETN VXX was launched in 2009, its inverse counterpart XIV in 2010 (it has since collapsed in February 2018 and been discontinued), the 2x levered version UVXY and the inverse ETF counterpart SVXY in 2011 as well as many other products.
• Options on VXX were introduced 2010, on UVXY and SVXY in 2012

• The Vix Futures Term Structure adds an additional source of return, it provides a considerable roll yield that is positive for a short position about 80% of the time and a part of which can be harvested.
• We can derive different trading signals with important market information contained in the term structure.
• Historical performance shows different return distributions for VIX derivatives and S&P 500 options – it will add diversification.
• The basic risk characteristics are the same however, therefore I trade volatility directly, but put it in the same portfolio allocation bucket as short option strategies.

New and Simplified Strategy Rules edited in May 2018
From recent experiences and additional analysis, I have decided to discontinue the use of volatility ETP and options and instead use Vix Futures and VIX options exclusively. It is much simpler with lower maintenance, avoids fees and transaction costs and should yield similar returns at lower exposure to volatility shocks.


Trading Signals

• Trend Filter: S&P Price > 275-day simple moving average
• Contango Filter: Be constantly short VIX Futures when VIX Futures Term Structure is in Contango
• Contango eg. vixcentral.com: VIX Future 1 < Future 2; if time to expiration of Future 1 is less then 5 days then Future 2 < Future 3
• Entry: Short 3 different lots of VIX contracts in staggered expiration months when contango > 0
• Exit: Exit all positions when contango < -1,5%
• Position size: Value of underlying is between 15%-25% NAV
• Roll contracts: Exit 1 month to expiration; enter new position 3 months further out
• Covered VIX option overlay: because of the predetermined exit we can sell VIX options to time it and harvest additional premium without additional risk and some hedging properties. Profits on the futures positions are limited to the option strike price and additional margin is required. I write 10 VIX Puts for each future with an expiration coinciding with the planned roll date of the future. The strike price is around the average current VIX lows of the last months. This makes most sense when the VIX has fallen close to these lows and we are waiting for the roll date harvesting roll yield – for me this has a strong behavioral component as it keeps me from rolling the futures early (the highest roll yield is in the front months).

The Original Rules as well as some additional analysis and alternative signals

Edit February 2018: An unprecedented volatility spike (VIX up 115.6% on February 5th) has triggered a discontinuation of XIV after falling more than 90% a day after the Term Structure signaled to exit all direct short volatility positions. This stresses the inherent danger and the importance of diligent risk control through moderate position sizes and strict stop losses.

• Momentum and trend: one can compare the price action of volatility ETP as a direct trading signal, using momentum (over a lookback period shorter than the usual 12 months – e.g. 80 days), but I prefer moving average trend indicators over different time frames in combination with term structure signals. As a basic trend filter – as throughout my portfolio – I use the market´s long term moving average, in this case the S&P 500´s. Vix tends to spike to extremes and cluster at high levels when the S&P 500 already trades below its 200 day moving average.
• Roll Yield: two simple methods can be used here: one can look at the current roll yield and the term structure curve visually at vixcentral.com or calculate the ratio between the Vix and its longer term cousin VXV: VIX/VXV for similar signals.
• The Volatility Risk Premium for Vix futures can be estimated directly with very good results, but this approach doesn’t make the same intuitive sense to me and I need to trust the signal to be able to stick to trading it.

• SVXY (XIV) and VXX medium-term trend following. The majority of strategies I found, trade both or one of these ETP. The aim is to reduce the drawdowns and volatility while participating in most of the upside of short volatility ETF SVXY (or near identical XIV). The main reason I use moving average cross over signals here is, that these generate slightly different entry times from the other sub-strategy signals.
  • Entry SVXY on moving average cross 9 day SVXY SMA above 60 day SVXY SMA and contango > 0 and S&P 500 > 275 day SMA; exit SVXY on moving average cross 9 day SVXY SMA below 60 day SVXY SMA or contango < -1,5%; position size 10%-15% NAV.
  • Long volatility VXX (to hedge tail risk): entry if contango < -3% and S&P 500 < 275 day SMA; position size 10%-15% NAV.
• Long UVXY put option. The advantage of being long puts is the built in maximum downside protection (in contrast to extremely risky short calls), while participating in the upside with leverage (about 4 times SVXY with the strategy parameters below). As the tendency of UVXY to lose value quickly is well known, the puts unfortunately demand a very steep premium. UVXY has to fall considerably for the put to become profitable, but nonetheless I consider this the best risk-reward strategy when looking at my backtests. Long puts have the highest exposure in the direct volatility trading part of my portfolio.
  • Long UVXY put rules: buy UVXY put when Vix term structure returns to contango > 0 after being in backwardation, expiration in about 5 months, strike OTM at last UVXY low; exit after two months (if UVXY should be at a relative high then, exit after three months); position size 5%-10% capital (put premium).
• Weekly UVXY short calls. UVXY calls command very high premiums mirroring their crazily high volatility (about 10%-15% premium for an ATM call 3 weeks to expiration at the moment) and they have the very strong directional UVXY bias working against them – resulting in a high probability trade when we are short. These calls lose more than 50% of their value (our profit) over the course of a week or two most of the time, but can spike up a multiple of that within a day or two every once in awhile. The aim is to create a high weekly income stream and avoid lethal spikes without getting whipsawed too much. A possibility is using call spreads for protection, but I found the trading costs getting very high in practice and decided to use a stop loss rule on a naked call instead. This leaves the strategy exposed to extreme black swan events and overnight gaps, a risk I mitigate through relatively small position sizes.
  • Selling UVXY calls weekly rules: the idea is to sell a call each friday to earn the premium, when conditions are favorable. I start the strategy, when the Vix term structure returns to contango > 0 after being in backwardation; sell one third position size call, three weeks to expiration, strike 2%-5% OTM, repeat each friday – the strategy will accumulate three open positions with staggered expiration dates; exit at expiration or at emergency stop loss at -200% or exit and stop trading the strategy when contango < -1,5%;  position size for 3 call positions 0,75%-1,5% NAV (call premium).
  • At the stop point (contango < -1,5%) a considerable amount of the earned premium will be given back – the rules aim to keep about 50% of the premium on average over time. If results deviate from this goal too much, I will change or discontinue the strategy. I posted a case study and update with several adjustments in the strategy´s rule following a strong double volatility spike in August 2017.
• Overall position size and exposure. In combination these strategies use a maximum of 15% to 25% of capital, but often only a part of the strategies is in the market simultaneously. Because of the inherent leverage in the options (the puts used are about 4 times, the calls about 12 times levered compared to SVXY), this will translate to a maximum exposure of approximately 50% to 100% NAV. I cap my maximum exposure at 75% NAV and only use full short call position sizes, when put and SVXY positions are so far in the green, that I can set a break even stop loss and remove them from the equation or if they are already closed.
  The position sizes on the high end are the maximum growth exposure (using the Kelly criterion) – I use a mid-level exposure at the moment. With a growing account size or growing skepticism about the market, I plan to move towards the position sizes at the low end or scale down even further.
• Strategy statistics: I closely observe real time trading data, but so far the strategies have only been active for a couple of months – with very good results. Backtesting the combination of strategies from the inception of UVXY (which excludes the very negative 2007-2009 period) yielded 39% annual return with 32% volatility, a Sharpe ratio of 1,2 and a maximum drawdown of -23% – in line with outside sources, but I expect future drawdowns to double easily given the volatility.