Featured - Posted on Sunday, January 18, 2015, 9:42 PM GMT +1


Jan Sunday 18

Timing (And Trading) Implied Volatility

The majority of readers will already be familiar with the fact that the CBOE Volatility Index® (VIX®) is not a tradable asset (it is just a number), and trading the VIX® in fact means trading its derivatives (futures) or even derivatives of derivatives (options on futures, ETFs/ETNs like XIV® – VelocityShares Daily Inverse VIX Short-Term ETN – and VXX – iPath® S&P 500 VIX Short-Term Futures™ ETN – ).

Due the mean reverting nature of the VIX® (e.g. when bottoming out in the 10-12 range, a sudden spike is much more likely than a further drop, and conversly after a sudden spike to extended levels, a (quick) drop and/or slow decrease to regular levels is just a question of time and much more likely than a further rise), timing and trading the VIX® would of course be much easier than timing and trading its derivatives (and derivatives of derivatives) where one has to take into account (and overcome) time to maturity (futures, options), time decay, risk premium (futures, options, ETFs/ETNs), roll yield (ETFs/ETNs), term structure (contango/backwardation of futures), seasonalities (e.g. FED announcement days, the last days before maturity, …), among others.

Some time ago MarketSci published an intersting article about timing (and trading) the VIX®Random Thoughts RE: Trading Volatility ETFs (Part 1) ), utilizing a 10-day EMA (Exponential Moving Average) and a 10-day SMA (Simple Moving Average), going long (selling short) the VIX® index at the close when the 10-day EMA of the VIX® closed under (over) the 10-day SMA. Expectably (selling short an asset which is always bottoming out in the 10 – 12 range) – in contrast to going long the XIV® (selling short volatility) – the short side of the trade was more or less treading water over the course of the time frame under review (since 1990) while the long side went straight up (low risk / high reward).

But as previously shown, with respect to a highly volatile asset like the VIX® , a 10-day moving average – even an EMA – is disadvantageous compared to a shorter-term moving average. And additionally – at least with respect to trading the Volatility Risk Premium Strategy – utilizing the CBOE Mid-Term Volatility Index ( VXMT® ) as a the repective trigger index instead of the VIX® may have some benefits again as well.

To make a long story short:

Image I shows the respective equity curves:

(1)  VIX® with 10d EMA vs. 10d SMA: blue line (complies to MarketSci’s posting)
(2)  VIX® with   3d EMA vs. 10d SMA: grey line
(3)  VIX® before 1/1/2008 , VXMT® after 1/1/2008 with   3d EMA vs. 10d SMA: red line
(4)  120% of VIX® | –20% of VXMT® with   3d EMA vs. 10d SMA: black line
       * just VIX® before 1/1/2008 , VXMT® after 1/1/2008
(5)  120% of VIX® | -20% of (VIX® + 10%) with   3d EMA vs. 10d SMA: green line
       * before 1/1/2008, VIX® had been increased by 10% in order to replicate the VXMT®


Image I – Total Equity Curve(s)
(01/01/1990 – present)


Some remarks are mandatory:

(1) Any additions/changes in the respective underlying are only related to the exponential moving average ( e.g. 120% of VIX® | –20% of VXMT® ). The 10-day SMA remains unchanged and is always based on the VIX® index.

(2) The blue line represents MarketSci’s 10-day EMA | 10-day SMA mean reversion strategy. The respective performance (hypothetically trading the VIX®) could’ve been easily boosted by utilizing a 3-day EMA instead of a 10-day EMA ( grey line ). Simply replacing the VIX® by the VXMT® index ( red line ) would be very disadvantageous (a least with respect to a mean reverting strategy) due to the fact that the VXMT® is regularly trading (significantly) above the VIX® index  ( contango ). Even better works a mixture of 120% VIX® minus 20% of VXMT® , regularly (artificially) reducing the index value ( black line ).

(3) This very simple mean reversion strategy works best when applying this kind of ‘mixture’ right from the start, means first of all simulating VXMT® index values in the simplest way by adding a constant 10% premium to VIX® index values before VXMT® index values are available (1/1/2008), and secondly applying the previously mentioned formula again ( 120% of VIX® | –20% of  (VIX® + 10%).

Image II shows the respective equity curves (long / short seperately) for MarketSci’s 10-day EMA | 10-day SMA (black / grey) and the 120% of VIX® | -20% of (VIX® + 10%) with 3d EMA vs. 10d SMA (green line) mean reversion strategy.


Image II – Total Equity Curve(s)
(01/01/1990 – present)


And last but not least – probably surprising the most – the respective Summation Index, simply representing the running total of net advances = raw quality of forecast (getting an index move right: +1 ; getting it wrong: -1). This image clearly shows that trading is NOT about being right or wrong (means just getting the direction of the move right), but all about making money (effectiviness and efficiency). MarketSci’s 10-day EMA | 10-day SMA (blue line) and the 120% of VIX® | -20% of (VIX® + 10%) with 3d EMA vs. 10d SMA (green line) mean reversion strategy are at equal level (at the end of the field !), but the latter strategy is doing things in an optimal way, being right when the VIX® index  moves big and losing small when being wrong (it ouperforms the 10-day EMA | 10-day SMA strategy by a factor of 1E+8) while the “VIX® before 1/1/2008 , VXMT® after 1/1/2008 with 3d EMA vs. 10d SMA” ( red line ) is at the top of the pack even after 1/1/2008, unfortunately winning small and losing big, depleting its net asset value since 1/1/2008 by 99.9%.


Image III – Summation Index
(01/01/1990 – 10/15/2014)


But how to take advantage of these findings will be subject to another posting. And may be some food for thought for your own analysis as well.

to be continued … (means more on this to come, stay tuned)


Have a profitable week,


Disclosure: I’am long/short XIV and long/short EURO STOXX 50 volatility futures.


XIV® VelocityShares Daily Inverse VIX Short-Term ETN
VXX® iPath® S&P 500 VIX Short-Term Futures™ ETN
VXMT® CBOE Mid-Term Volatility Index



Remarks: Due to their conceptual scope – and if not explicitly stated otherwise – , all models/setups/strategies do not account for slippage, fees and transaction costs, do not account for return on cash and/or interest on margin, do not use position sizing (e.g. Kelly, optimal f) – they’re always ‘all in‘ – , do not use leverage (e.g. leveraged ETFs), do not utilize any kind of abnormal market filter (e.g. during market phases with extremely elevated volatility), do not use intraday buy/sell stops (end-of-day prices only), and models/setups/strategies are not ‘adaptive‘ (do not adjust to the ongoing changes in market conditions like bull and bear markets). Index data (e.g. S&P 500 cash index) does not account for dividend and cash payments.



The information on this site is provided for statistical and informational purposes only. Nothing herein should be interpreted or regarded as personalized investment advice or to state or imply that past results are an indication of future performance. The author of this website is not a licensed financial advisor and will not accept liability for any loss or damage, including without limitation to, any loss of profit, which may arise directly or indirectly from use of or reliance on the content of this website(s). Under no circumstances does this information represent an advice or recommendation to buy, sell or hold any security.

I may or may not hold positions for myself, my family and/or clients in the securities mentioned here. Actions may have been taken before or after information is presented, and any opinions expressed in this site are subject to change without notice.

(Data courtesy of MetaStock and Pinnacle Data Corp., and for data import, testing, surveys and statistics I use MATLAB from MathWorks)

Comments (31)


  1. Mike says:

    Greetings (to the author of trading the odds),

    I really appreciate (am genuinely thankful for) the work you do (the studies you accomplish) on Trading the Odds (this volatility website). I just have one simple question (request), would it be possible (or even probable, since we trade in probabilities :) for you (or whoever publishes your work) to drop the majority (at least a good half or more) of the parenthesis () that you write (or type) in?

    It’s just hard to read is all, and the majority of the writing could be just fine without all the parenthesis.

  2. Ilya Kipnis says:

    From my experience, the VIX is far, far different than its derivatives. As I posted a while back, the VIX has some dynamics unique to the fact that it is not an instrument that can be traded.



    The VIX has a weekend premium. This does not reflect to its derivatives, for instance.

    I’ll wait to see where this goes.


    • Marco says:

      yes was wondering the same. As you point out yourself you have to trade the VIX via its derivatives – so why analyse the performance of buying and selling the VIX (spot) Index?
      Supposedly the equity graph would look very differently if you trade the signal via VIX futures or the usual ETNs (XIV/VXX etc) – how comes you didn’t analyse that?

      • TradingTheOdds says:


        timing and trading the VIX efficiently (being right when the VIX index moves big and losing small when being wrong) is definitely helpful trading its derivatives (futures and XIV/VXX) in order to overcome time decay, roll yield, term structure … (whatever works against going long the VIX).

        With respect to your second question: Did you notice the last para. : ‘And may be some food for thought for your own analysis as well. to be continued … (means more on this to come, stay tuned)’


        • Marco says:

          okay I will wait for the follow though and see what you make of this, but I remember a similar analysis that worked great for ‘trading’ the VIX index but when you bought/sold the front month future or XIV/VXX rather than the VIX the return went from a few 1000% to single digit returns. So it’s relatively easy to time the VIX index but a lot harder to make money from the right timing and trading its derivatives (So its easy to predict that VIX will go up once it hits 10, but as the future trades at say 14 and by the time the VIX went back to 11.5 (you timed the VIX right) the future dropped to 13 and you lost money…) you know all of that of course but that’s why I question the value of that analysis but please prove me wrong with the next post :)

          • TradingTheOdds says:


            to be totally clear: I neither own a crystal ball nor would I post (in detail) about something that gives me a real competitive edge in the market (if you know anybody who does – for free -, please let me know). If you never here about timing the VIX again, its either because there is nothing to write home about, or … (you know).

            If you find something new and interesting here, I recommend to build upon and do some math and analysis yourself. Please share your results as I do. I doubt that you’ll find the jack of all trades device (for free) anywhere on the web. The blog is all about ideas, strategies, statistics, not about customized solutions ready to trade.


          • Marco says:

            this is as a reply to the below (for some reason can’t reply there)

            Hi Frank,

            I appreciate the work you do at this page and that you share your insights and ideas. So my question ‘Why do you analyse and try timing the VIX?’ – was just that – a genuine question. I did not try to imply you are stupid doing that or are wasting your or anyone elses time. I was just wondering how is does help you (or me) to trade VIX derivatives better. You partially answered that by saying by knowing what the VIX does you can predict better what VIX derivatives do. My mistake was to take that as a stand alone analysis rather than a first step – so I hold on to see whether you publish a follow up or not (if you found nothing worht reporting or found the holy grail of VIX trading and rather keep that for yourself)

          • TradingTheOdds says:


            again: The blog is about ideas, concepts, strategies, and so on. I love to see such analysis anywhere else on the blogospehre (e.g. https://marketsci.wordpress.com/, http://volatilitymadesimple.com/blog/ ), because it allows me to build off of their ideas in order to come up with an(other) unique edge in the markets. I don’t have a solution to every topic/problem/question I’am blogging about, but may be another blogger/reader has willing to share.

            I’am sure that if one is able to forecast big moves in the VIX index with a high accuracy, that will definitely help trading VIX derivatives as well.


  3. […] Timing (And Trading) Implied Volatility [Trading the Odds] The majority of readers will already be familiar with the fact that the CBOE Volatility Index? (VIX?) is not a tradable asset (it is just a number), and trading the VIX? in fact means trading its derivatives (f… […]

  4. Fepson says:

    Hi Frank,

    Isn’t it going long the VIX index when its 10-day EMA close “over” its 10-day SMA instead of “under”, m’I correct ?

    Thank’s for sharing,


    • TradingTheOdds says:


      no. Go long (short) the VIX when the x-day EMA of the VIX closed under (over) the 10-day SMA is absolutely correct. The 10-day EMA is moving faster than the 10-day SMA, leave alone the 3-day EMA. Its a mean reversion strategy (means ‘buying’ when the VIX is ‘oversold’, ‘shorting’ when the VIX is ‘overbought’), not a trend following / momentum strategy.


  5. Marco says:

    As I have been a great fan of Marketsci’s article, I enjoyed this so much that I look forward for the followings. I think we should focus on VIX and its etps differences to find the best filter to trade them.

  6. […] Timing (And Trading) Implied Volatility | TRADING THE ODDS. […]

  7. Jacky Chen says:

    It seems to be a nice idea but if you use the old VXO as a proxy and extend the data to 1987 you would simply be wiped out overnight. It is a rare event indeed. But never forget that you could be destroyed all of a sudden when betting on the mean reversion of volatility:-)

    • Jacky Chen says:

      In the original Marketsci setting I think you would go short VIX as there was an big increase in VIX before Monday if my codes are right. If you could short the actual VIX with a leverage of 1 you would be wiped out. If you use SVXY assuming a 60% VIX beta=>VXX would more than double and SVXY return=1-VXX return=>It would go to zero. If you short VIX futures you would have a negative asset. You would not have anything left before the 10/21 rebound and it is just like what happened to some long USDCHF guys last week.

      I am not saying it is not an useful idea. But it needs to be played with caution and a small bet with SVXY might be the best choice as you will not lose more than your initial bet.

      • TradingTheOdds says:


        you’re correct, the strategy’s net asset value went from $0.36 to $-0.67 on 10/19/1987. Assuming that you’d go on even with a negative stack (on credit), a year later you would’ve been back to even, and assumed again that you start all over again with $1 at the end of 1989, your current net asset value would now be at $9E15.

        Timing and trading the VIX is for statistical purposes only in order to proove if it provides an edge trading XIV and VXX. When trading VXX/XIV in a real account one has to consider a lot more: Position sizing, money management, … It would be possible as well to bet a maximum of $10,000 (only) and put aside any gains into a separate (money market) account.


  8. MikeA says:

    Very interesting, but I must confess I am confused by some of your nomenclature. Can you explain this:

    120% of VIX | -20% of VXMT

    Are you deriving a new metric that is made up of the VIX and VXMT and applying the EMA and SMA on that derived metric?

    What is the idea behind this derived metric?

    • TradingTheOdds says:


      simply apply the formula presented: 120% VIX® minus 20% of VXMT® (e.g. VIX at 20, VXMT at 22 : 120%x 20 -20%x 22 = 19.60).

      Additionally see:
      ‘(1) Any additions/changes in the respective underlying are only related to the exponential moving average ( e.g. 120% of VIX® | -20% of VXMT® ). The 10-day SMA remains unchanged and is always based on the VIX® index.’

      At the end the VIX is artificially reduced when in contango (VIX < VXMT) and increased when in backwardation (VIX > VXMT), intensifying a mean reversion behaviour.


  9. Dewey says:

    I agree with previous post about using actual tradeable symbols such as VXX, XIV, ZIV, VXMT to produce a real backtested product over the past three years when most of these ETP’s actually existed. Then so your best to extrapolate that data back to 1987 or earlier.

    I still don’t understand what you were trying to explain to Mike, “simply apply the formula presented: 120% VIX® minus 20% of VXMT® (e.g. VIX at 20, VXMT at 22 : 120%x 20 -20%x 22 = 19.60).” What does that mean? I assumed you were buying some type of leverage of VIX at 1.2 and shorting VXMT by 20%? The equation and intent is not clear.

    Even your disclosure is confusing. What does, ” I am long/short XIV…” mean?

    I truly can’t wait to see your analysis using tradeable exchange traded products.

    • TradingTheOdds says:

      For every day, calculate:
      After 1/1/2008: (1.20 * VIX – 0.20 * VXMT).
      Until 12/31/2007: (1.20 * VIX – 0.20 * (VIX * 1.1))

      Then take the 3-day exponential moving average of these values. This represents the 3-day-EMA of …
      The 10-day simple moving average (SMA) of the VIX should be clear.

      And if you take a look at the images(s): Tradable asset is the VIX (hypothetical results, of course, the VIX is NOT a tradable asset).

      The “Disclosure” is for legal purposes only: It means I’am either long or short (or both, if I’am in a spread) …

  10. tapesense says:

    Another great post. I am curious what monte carlo analysis might offer, e.g. to see what an abnormal grouping of negative trades might look like. Is there a 1 in 10 possibility of a 50% drawdown, or 1 in 100? Also, have you found any results using the futures contract or etfs based on it? Thanks much

  11. Future says:

    I am believe that Today ia a Buy in vxx,vxz,vix have a divergence with them,but this is not reason.
    The cycle natural active i have study,this is private,lt is the reason,this happen medium 2 to month.
    I am buy 23 march vxx but here can go second buy,the Buy cancel if today have a strong up in vxx.
    27 march and 10 april are points Big probably inflexión up/down vxx,vix,10 april is near mercury conjuction aspect.
    Remember 16 october or 17 december 2014,the same.

  12. Florent says:

    Hi Franck, I played with some R code and I’ve the same result (green line) but I can’t understand how returns can be so different when trading VXX and XIV. Normally when VIX goes up VXX goes up and vice versa. Any idea ?

    • TradingTheOdds says:


      what do you mean with “how returns can be so different” (long XIV / short VXX doesn’t deliver same returns ?) ?


      • Florent says:

        I used the 120% VIX -20% VXMT to generate signals but buy and sell with XIV and VXX. Returns are horrible. For some reason the strategy turns from a exeptional strategy to a non profitable.


  13. Xavier says:

    I think that, compared to a VRP strategy (e.g. 5EMA of (VIX-2dHVOL SPX)), these EMA/SMA crossover strategies require too many trades, thus increasing transaction costs.
    Would be interested though, to see the number of trades and underwater curves.
    Best to you.

  14. Luke says:

    I have tried to match your results using the formulas and during 2008 my balance goes negative. I would have been short at the close of 09/02/2008 until the signal switched to long on 10/21/2008. I am using closing prices to caclulate the 3 day ema and your formulas. Did you have a severe drawdown during this timeframe as well?

    Thank you for your help,


    • TradingTheOdds says:


      if you take a look at the respective chart (comparison of Net Asset Values, in this event the gree line) you can see that the strategy faced a severe drawdown in 2008. It takes too long to replicate formulas and stretegys (I don’t save everything I ever tried out), but yes, you will probably be right.


  15. Richard says:

    Xavier (and all reading)…

    I found this study fascinating. I reproduced it and have been experimenting.

    The drawdowns are horrific. The problem comes with a big spike… sometimes a short trade is placed just before a spike, and it doesn’t close until after the spike peaks. The reverse happens right after a big spike, the system puts you into a long trade that you ride all the way down.

    Can you think of any way to reduce the draw-downs?


    • TradingTheOdds says:


      due to their conceptual scope, all models/setups/strategies do not use position sizing (e.g. Kelly, optimal f) and/or money management – they’re always ‘all in‘ – , do not use leverage (e.g. leveraged ETFs), do not utilize any kind of abnormal market filter (e.g. during market phases with extremely elevated volatility), do not use intraday buy/sell stops (end-of-day prices only), and models/setups/strategies are not ‘adaptive‘ (do not adjust to the ongoing changes in market conditions like bull and bear markets).

      An application of at least one (or a combination of) those measurements listed above should help. Still on my list of open issues …


Leave a Reply

Your email address will not be published. Required fields are marked *