Seeking Alpha

Marc Cohn's  Instablog

Marc Cohn
Send Message
I am a patent litigator with a background in physics and electronics. I enjoy studying quantitative, rules-based investment methods through rigorous backtesting and numerical analysis. I believe that patterns exist in the market that benefit trading -- the challenge is finding them!
  • New Options Trading System Signals (CAGR Over 500%??) 21 comments
    Jul 14, 2014 6:08 PM | about stocks: VXX, SSO

    I know 500%, whatever. But the back testing is what it is. No guarantee of future performance, of course... but let's discuss. This is more of a learning exercise than anything else. A "let's see what happens" discussion.

    This is a system for trading VXX options that has back tested remarkably well since VXX launched in 2009. I want to post the signals here going forward so I can establish the system in real time.

    (For those of you who don't know, VXX is an exchange-traded note related to the short term futures of the VIX volatility index. If you didn't already know what the VXX or the VIX is, I would caution you against trading this system until you learn more about volatility trading.)

    The principle of this system is simple: every Friday, you check to see if the VXX is trading relatively high compared to its recent performance. If so, you buy weekly VXX put options at the money only if the option is not too expensive. Roll the option until the VXX has fallen to a relatively low level or until the options are too expensive. In between these trades, we invest in SSO.

    The formulae for determining when the VXX is high and low and when the options are expensive are my secret sauce. Also secret sauce is the strike price of the options and how much of the portfolio to invest in the options (i.e. the money management side).

    I'm sure many of you will be thinking that I've done a lot of curve fitting here. But I know the dangers of curve fitting and I do my best to steer clear. The formulae for the VXX valuations are very simple. Some of you may even be able to figure them out. Also, when the parameters are changed (hint, there's an EMA in there somewhere), the returns are still outrageous, even if not optimal.

    Using some middle-of-the road parameters, the returns since 1/30/2009 are astounding:

    Total13758.61
    CAGR572.54%
    Stdev144.7%
    Sharpe3.96
    Max DD44.80%
    CAGR/DD12.780
    Linearity35.450%
    Growth R16.150
    Win %61.86%
    Trades97

    (Linearity and growth R are metrics described in some of my other blog posts and articles.)

    These are not the best values that can be obtained, but the desire to avoid curve fitting led me to follow the simplest system that provided the most robust returns. Did I run a multi-dim analysis on this? I did not. Through trial and error, I tried to find a set of parameters whereby the total returns and Sharpe ratio were least affected by changes in those parameters.

    The total returns using my baseline parameters are shown on a log graph below. Note this is not percent... that is 10,000 times the original investment.

    (click to enlarge)

    Sure, that's all a little nuts, but the numbers are what they are.

    Anyway, I will be posting signals here and we can all see what happens.

    Some recent history:

    The system had us buy VXX puts on 3/28/2014 at the close (43.55). We sold them a week later at the close (41.84) for a profit of around 4%. The numbers in ( ) above reflect the VXX price -- the options themselves obviously were different, on the order of about $1.50 per contract.

    The system had us buy VXX puts on 4/11/2014 at the close (44.75) and a week later VXX was at 41.71. We rolled the puts for a tidy profit of 14% and sold them another week later at the money for a loss of about 9%.

    In between these two trades, we held SSO for a loss of 3.2%.

    Since the last VXX trade closed on 4/25/2014, we have been long SSO. This open position is currently seeing a gain of 12.5% (including today).

    Thus, since 3/28/2014, we have seen a net gain of about 18% (counting today).

    Since 1/2/2014, the system is up about 65%. There have been 6 VXX trades this year of: 16% (1/31), 14% (2/7), 14% (4/11), 11% (3/14), 4% (3/28), and -9% (4/17). That's a net gain of 58%. The rest is from being in SSO between VXX positions.

    Currently the VXX is well below the trigger value, so we may have some more time yet before another VXX trade is signaled. In the meantime we hold SSO and pray.

    Stay tuned for more signals and discussion...

    Disclosure: The author has no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

    Stocks: VXX, SSO
Back To Marc Cohn's Instablog HomePage »

Instablogs are blogs which are instantly set up and networked within the Seeking Alpha community. Instablog posts are not selected, edited or screened by Seeking Alpha editors, in contrast to contributors' articles.

Comments (21)
Track new comments
  • Algyros
    , contributor
    Comments (101) | Send Message
     
    What can I say that isn't a monumental understatement.

     

    To put your ideas in perspective, do you have any sense of what kinds of returns this system would generate if, instead of options, one were to simply invest in a VXX short when the signal came?

     

    And, needless to say, I'm eager to learn more.
    14 Jul, 06:43 PM Reply Like
  • Marc Cohn
    , contributor
    Comments (87) | Send Message
     
    Author’s reply » Doesn't work for a short. The good thing about the puts is that they are like lottery tickets. For a price, you get to win big or else you only lose what you put in. If you short VXX, you can have enormous downsides suddenly. If you try to prevent that with a stop loss, you will often exit the position and miss you on a recovery after a big fall. With a put, you will stay in the position regardless of whether the price goes against you and past your strike price, then if the price comes back in your favor, you remain in the position.
    15 Jul, 08:38 AM Reply Like
  • tmdoherty
    , contributor
    Comments (257) | Send Message
     
    Hi Marc,

     

    I was wondering the same thing. Hard to judge a system that trades puts, since the spreads and slippage and other costs can be considerable.

     

    Do you employ any stop-loss strategy? How would the system deal with big gap ups in VXX due to overnight global chaos of some sort?
    14 Jul, 07:49 PM Reply Like
  • Marc Cohn
    , contributor
    Comments (87) | Send Message
     
    Author’s reply » No need for a stop-loss when you're long a put -- the put itself is the stop loss. If the trade goes against you, then you lose the entire put, but no more than that.

     

    Obviously, you do not put 100% of your principle into the puts or you would lose everything on one bad trade. There are money-management techniques applied that dictate how much to invest each time.

     

    As for SSO, I didn't build a stop-loss into that. I suppose you could put in some sort of black swan loss preventing, like a 20% stop loss. The system as backtested did not have that. As SSO begins to fall, VXX rises and then the system will trigger you to sell SSO and buy VXX puts.
    15 Jul, 08:26 AM Reply Like
  • Left Banker
    , contributor
    Comments (2128) | Send Message
     
    Me too on VXX short instead.

     

    VXX can bite you on the butt. Big time. Overnight. I think I'd want some insurance here, but it will cost.

     

    Consider: Buy SVXY. Buy protective puts to limit losses to some tolerance level. Not 500%, but ok. And by "ok," I mean "OK!"
    14 Jul, 09:44 PM Reply Like
  • OwnPuts
    , contributor
    Comments (17) | Send Message
     
    Here is some more historical data:
    http://bit.ly/1mPJA7i
    15 Jul, 12:36 AM Reply Like
  • Marc Cohn
    , contributor
    Comments (87) | Send Message
     
    Author’s reply » Thanks, this is very helpful. I will see how this goes.
    15 Jul, 09:04 AM Reply Like
  • Marc Cohn
    , contributor
    Comments (87) | Send Message
     
    Author’s reply » I have used the data at the link you provided and the results still look promising. The pre-2009 data is not "historical" as you describe but was calculated using some unknown formula based on the VIX short term futures contracts.

     

    Using this calculated data from 2004 to 2009, the results were CAGR 37% with a drawdown of 62%. After 2009, the results are as described in the post above.

     

    The log returns for both the pre-2009 and post-2009 data are rather uniform. In other words, the log of returns looks like a boomerang, with a relatively straight line from 2004 to 2009, then another relatively straight line from 2009 on, with the latter being at a steeper angle. In light of this result, I am skeptical that the pre-2009 generated data in the link you provided is an accurate reflection of VXX pre-2009.

     

    Below are the returns of the system by quarter:

     

    Quarter Total Percent
    4/1/2004 1.03
    7/1/2004 1.12 8.27%
    10/1/2004 1.27 13.80%
    1/1/2005 1.32 3.39%
    4/1/2005 0.93 -29.03%
    7/1/2005 1.50 60.36%
    10/1/2005 1.62 8.43%
    1/1/2006 1.76 8.23%
    4/1/2006 1.81 2.80%
    7/1/2006 1.63 -9.93%
    10/1/2006 2.23 36.98%
    1/1/2007 2.50 12.01%
    4/1/2007 3.53 41.37%
    7/1/2007 3.16 -10.39%
    10/1/2007 3.58 13.21%
    1/1/2008 3.42 -4.43%
    4/1/2008 3.16 -7.70%
    7/1/2008 4.31 36.58%
    10/1/2008 2.47 -42.64%
    1/1/2009 4.07 64.64%
    4/1/2009 6.13 50.56%
    7/1/2009 7.80 27.30%
    10/1/2009 13.03 67.02%
    1/1/2010 15.16 16.31%
    4/1/2010 22.41 47.81%
    7/1/2010 50.60 125.84%
    10/1/2010 107.35 112.15%
    1/1/2011 157.44 46.65%
    4/1/2011 188.99 20.04%
    7/1/2011 269.95 42.84%
    10/1/2011 215.18 -20.29%
    1/1/2012 783.55 264.14%
    4/1/2012 1181.20 50.75%
    7/1/2012 3006.49 154.53%
    10/1/2012 5977.08 98.81%
    1/1/2013 7525.82 25.91%
    4/1/2013 20858.09 177.15%
    7/1/2013 19419.70 -6.90%
    10/1/2013 28110.10 44.75%
    1/1/2014 40181.14 42.94%
    4/1/2014 56669.53 41.04%
    7/1/2014 67610.39 19.31%
    15 Jul, 12:07 PM Reply Like
  • OwnPuts
    , contributor
    Comments (17) | Send Message
     
    Here are two other attempts to create simulated data prior to etf initiation:
    http://bit.ly/1nEex0Y

     

    http://bit.ly/1lNYl8P
    15 Jul, 01:42 PM Reply Like
  • Marc Cohn
    , contributor
    Comments (87) | Send Message
     
    Author’s reply » With the data in the first set, the results are:

     

    Total 5776609.03
    CAGR 374.42%
    Stdev 68.2%
    Sharpe 5.49
    Max DD 51.64%
    CAGR/DD 7.250
    Linearity 116.892%
    Growth R 3.203
    Win % 53.48%
    Trades 187

     

    And the results by quarter are:

     

    TOTAL % Per Quarter Trades
    4/1/2004 1.03
    7/1/2004 1.31 27.16% 2
    10/1/2004 1.29 -1.40% 0
    1/1/2005 1.35 3.96% 0
    4/1/2005 1.22 -9.05% 5
    7/1/2005 2.48 102.82% 8
    10/1/2005 2.77 11.60% 2
    1/1/2006 3.04 9.81% 5
    4/1/2006 3.13 2.80% 0
    7/1/2006 3.52 12.69% 7
    10/1/2006 5.18 47.11% 3
    1/1/2007 5.81 12.01% 0
    4/1/2007 9.69 66.87% 4
    7/1/2007 9.85 1.61% 9
    10/1/2007 11.67 18.56% 13
    1/1/2008 14.63 25.32% 7
    4/1/2008 16.02 9.52% 9
    7/1/2008 25.72 60.54% 4
    10/1/2008 21.22 -17.50% 7
    1/1/2009 33.80 59.29% 12
    4/1/2009 71.81 112.44% 6
    7/1/2009 97.86 36.28% 3
    10/1/2009 180.04 83.98% 2
    1/1/2010 313.73 74.26% 3
    4/1/2010 511.42 63.01% 3
    7/1/2010 1548.41 202.76% 7
    10/1/2010 4133.83 166.97% 5
    1/1/2011 6453.91 56.12% 1
    4/1/2011 5849.39 -9.37% 4
    7/1/2011 9529.31 62.91% 2
    10/1/2011 6865.26 -27.96% 12
    1/1/2012 17145.55 149.74% 8
    4/1/2012 29538.57 72.28% 2
    7/1/2012 103550.33 250.56% 7
    10/1/2012 224806.65 117.10% 2
    1/1/2013 356845.95 58.73% 5
    4/1/2013 1068780.59 199.51% 3
    7/1/2013 1118203.07 4.62% 6
    10/1/2013 1870797.78 67.30% 4
    1/1/2014 2982380.66 59.42% 5
    4/1/2014 4589136.47 53.87% 3
    7/1/2014 5756527.55 25.44% 3
    15 Jul, 02:17 PM Reply Like
  • MCFA
    , contributor
    Comments (11) | Send Message
     
    40% draw-down? You are brave!
    15 Jul, 12:52 AM Reply Like
  • Marc Cohn
    , contributor
    Comments (87) | Send Message
     
    Author’s reply » 40% is less than a simple S&P 500 ETF in the last 10 years. Drawdown for SPY was over 55% in late 2008-early 2009. There's no doubt the system as backtested is risky. But the price of that risk is not high, given the backtested returns.

     

    (I use the qualifier "backtested" returns because I dont want to suggest that these returns are sure to continue.)
    15 Jul, 08:24 AM Reply Like
  • Algyros
    , contributor
    Comments (101) | Send Message
     
    Needless to say, I find this idea intriguing. But, I'm having a hard time getting my frequently obtuse mind around it. Both SSO and a VXX put correlate closely to a long SPY environment, correct? If I am indeed right about this, then could you explain why you invest in SSO when you're not in a VXX put?

     

    Again, thank you for sharing this idea. Will you be, in fact, posting signals when they arise?
    16 Jul, 09:12 AM Reply Like
  • olmendreef
    , contributor
    Comments (19) | Send Message
     
    Hi Marc,
    There's a lot of 'secret sauce' in this system.
    Care to share with fellow investors?
    BR.
    17 Jul, 05:48 PM Reply Like
  • olmendreef
    , contributor
    Comments (19) | Send Message
     
    Dear all,
    I found something that looks interesting:
    http://bit.ly/1qwo11E
    In short: short VXX (or buy XIV) if VIX/VXV<0.92, buy VXX if ratio is >1.02. Stay in cash in between.
    Any comments are welcome.
    21 Jul, 06:00 AM Reply Like
  • zydecosailuh
    , contributor
    Comments (37) | Send Message
     
    I think you will want to look at this...
    http://seekingalpha.co...

     

    One of the more interesting issues related to the VIX:VXV ratio is how that ratio has moved up and down over the course of various volatility regimes. In the first two years of the ratio (2007 and 2008), the average (mean) ratio was 0.97 and 1.02; during the past two years, however, the average (mean) ratio has fallen all the way down to 0.87 and 0.90.

     

    For about 1 ½ years following the launch of VXV, using absolute levels in the VIX:VXV ratio as bullish and bearish signals worked almost perfectly. One only has to look at VIX:VXV Ratio Moving Toward Bearish Zone to understand why it quickly gained so many fans. When the 2008 financial crisis hit and in the aftermath of the crisis, the VIX:VXV ratio began to look like just another broken indicator. Since that time, the VIX:VXV ratio has continued to be a very useful indicator, yet its value has been greater when studied in relative terms than absolute terms. Old rules, such as increasing long equity exposure when VIX:VXV fell to 0.92 or 0.90, proved to be of little help when the average of the ratio hovered around 0.90 instead of 1.00.

     

    During the last two years, the VIX futures term structure has departed significantly from historical norms, as I have demonstrated in the likes of The 2012 VIX Futures Term Structure as an Outlier. Frankly, the 2013 version of the VIX futures term structure looks a lot more like the outlier 2012 data set than any previous year.
    6 Aug, 01:04 PM Reply Like
  • Viperace
    , contributor
    Comments (2) | Send Message
     
    Hi,

     

    Nice sharing.
    Do your system work for trading the Market Open instead of close? Ie, market close yesterday, you compute the whether to enter, then you enter the next day open.

     

    I recall EP Chan saying option is cheaper intraday than in close, perhaps u could consider that
    5 Aug, 09:25 PM Reply Like
  • nzbryant
    , contributor
    Comments (4) | Send Message
     
    Marc

     

    Great looking system. What program do you use to backtest put options on VXX, and where do you get data? (there must be so much data for every strike and expiry).

     

    I use Wealth Lab but dont know how I would apply it to this.

     

    Thanks
    11 Aug, 09:20 AM Reply Like
  • Marc Cohn
    , contributor
    Comments (87) | Send Message
     
    Author’s reply » Sorry, it has been so long! My day job has been keeping me very busy and it will continue to do so until late September.

     

    Obviously, this is not real time, but the system said to buy VXX puts on July 18 (Friday) and sell a week later. It would've been a loss of 10%. Ugh.

     

    Then there was a week off.

     

    Then there was a signal to buy VXX puts on Aug 1 and sell a week later. This was a 6% loss.

     

    Then there was a signal to buy VXX on Aug 8 and sell a week later. This was a gain of 21%. Huge.

     

    We were also supposed to but VXX puts last Friday, Aug 15. That trade is showing a profit but it aint over til it's over (this Friday.)

     

    All in all, it's not been great in July-August. These trades together net: .9 * .94 * 1.21 = 2.4%

     

    Stay tuned.
    18 Aug, 03:25 PM Reply Like
  • Allan Harris
    , contributor
    Comments (147) | Send Message
     
    VXX is on a Sell signal from August 11th. Although trading options on the Sells have been profitable, the REAL money will be made when (and if) the market goes through a 20% or more bear market. Outstanding gains from near-term calls will be made (been there, done that). We are treading water, but when the bears take control, the fireworks begin. Patience.

     

    http://seekingalpha.co...
    25 Aug, 12:23 PM Reply Like
  • nzbryant
    , contributor
    Comments (4) | Send Message
     
    Great looking system. What program do you use to backtest put options on VXX, and where do you get data? (there must be so much data for every strike and expiry).

     

    I use Wealth Lab but dont know how I would apply it to this.

     

    Thanks
    26 Aug, 08:59 AM Reply Like
Full index of posts »
Latest Followers

StockTalks

More »

Latest Comments


Posts by Themes
Instablogs are Seeking Alpha's free blogging platform customized for finance, with instant set up and exposure to millions of readers interested in the financial markets. Publish your own instablog in minutes.