Karen the Supertrader - TastyTrade Hybrid Experiment

Discussion in 'Journals' started by Sweet Bobby, May 18, 2016.

  1. isotope


    Several years ago, James Bittman of CBOE proposed his 2-step strategy. The backbone of the strategy is based on the following probabilities:-
    Prob of touching 1 Std Dev* (up or down) 54%
    Prob of touch 1/2 Std Dev* (up or down) 99%
    Touch 1/4 Std Dev* > 99%
    Touch down 1/4 after up 1/4 is touched* 34%
    Touch down 1/2 after up 1/2 is touched* 22%

    * Probability of touching any time during the period
    Note: probabilities are independent of time frame and level of volatility.


    The game plan:
    1. determine the time frame, e.g. Karen's favorite of 56 DTE
    2. use yesterday closing VIX and compute the volatility for 56 days: sqrt(56/365)*VIX
    3. compute from today's SPX opening price, the expected ranges i.e. -0.5SD -0.25SD +0.25SD +0.5SD (Std Deviation) for 56 days
    4. now wait for SPX to touch either -0.25SD or +0.25SD from the opening price (may take more than 1 day)
    5. if -0.25SD is touched, we assume a bearish trend, then we sell OTM calls or calls spreads with the short strike at +0.5SD
    6. if +0.25SD is touched, we assume a bullish trend and sell OTM puts or puts spread with the short strike at -0.5SD
    7. if we were wrong in our trend hypothesis, and SPX turns around, then close the position when the opposite 1/4SD is touched. Otherwise, hold the position until expiration.
    According to the probabilities listed above, we would average 2 home runs out of every 3 trades. Bittman back-tested his strategy using a 7DTE time frame over 41 weekly expiration cycles. The empirical results were consistent with the theoretical statistical model. Somebody else tested the strategy over various time frames on a Monte Carlo engine, and yielded similar results. There is another guy (forgot his name) who started an algorithmic trading company using an automated version of this strategy.

    The chartists and technicians severely criticize this strategy, citing the low profit margin in selling calls when the underlying price is falling and selling puts when price is rising. What they presume is their market timing with charts is superior to the mythical power of the central limit theorem.

    The take away of this strategy is that we can mechanically sell premium in a much wider range of conditions. However, I wouldn't play this strategy if VIX were below 12 or above 28. Neither would I pick the 56DTE time frame for real trades. Writing weekly options are more suitable due to their higher efficiency in theta collection. http://ir.cboe.com/press-releases/2016/27-01-2016.aspx
    Last edited: May 28, 2016
    #81     May 28, 2016
    Truth_ likes this.
  2. This is very interesting. Thanks for sharing. I'm going to study this more.
    #82     May 28, 2016
  3. Tastytrade did some studies that found closing trades with 21 DTE yielded similar results to closing trades at 50% of max profit. This was based on opening trades at 45 DTE. So the trades would have a duration of 24 DTE. Since some of my trades are in excess of 45 DTE, I decided to close them after 24 days instead of close them with 21 days remaining to give them enough time to decay, but at the same time closing them to take risk off the table so that I can put new trades on.

    I initially started this experiment thinking that I would sell Two puts to one call every week. I soon found it hard to do this on up days. So I adjusted my plan to look at my Greeks as well as daily direction. Generally on down days I will sell more puts to calls and on up days I will sell more calls than puts. Ultimately, the Greeks will control my decisions. I'm usually selling puts and calls together because I want to be on both sides of the market. Yesterday, however, I only sold puts because my risk to the upside is overly maxed out. I sold the puts even though the market was up. Keep in mind the puts have a 5% chance of being in the money and my Greeks indicated that my portfolio could withstand the addition of more puts.
    #83     May 28, 2016
  4. Conduit, please feel free to follow the account at

    Your assumption that the account will blow up is in error. The account is maxed out on the up side. A 15% up move would hurt me greatly. Since markets don't crash up, I will have plenty of time to close trades and adjust positions. Based on my observations, I am not comfortable putting on any more calls until I close some of my current trades.

    Now, if the market opens down big and volatility increases will my account be blown out? I have stress the account (as best I can) to withstand a 20% down move with a substantial increase in volatility if I make no adjustments to the account.

    In my opinion, everything is perfectly safe.

    Let me know if you have other questions.

    #84     May 28, 2016
  5. conduit


    In this post of yours


    you stated you have short vega exposure of -786. Looking at the table you pasted above and backing out an initial account size of around 20k a 1 vol point move against you will set your account back by around 4%. I did not mean to say that one large move might blow out your entire account (my apologies if I miscommunicated) but nonetheless 2-3 vol point moves are not unheard of at all after a weekend "special event". That would set you back at least 12% if not more depending on your gamma profile. Sounds more like a very risky approach to trading in exchange for collecting a bit of premium here and there. At the very least I would say no professional fund allocator would find such risk/reward acceptable. Please correct if any of my above numbers seem off. Something with your cum annualized relative return does not add up by the way, initial $86 depicting a cum annual return of 1.05% and $2308 (row8) depicting 11.7% seems hard to reconcile...

    Last edited: May 28, 2016
    #85     May 28, 2016
  6. The 11.7% return is a projection on my realized profit/loss. The formula makes an assumption that the profit/loss already obtained will continue thelroughout the year. I take the realized daily profit/loss and multiply by 365 days. Then, I divide that number by the beginning balance in the account. That gives me a projected annualized profit/loss as if I had been trading this account from January 1.

    I love your comment about unacceptable risk and that it would not be acceptable to any professional fund allocator. I'm just a simple guy testing a little strategy. It helps me to learn as I think through my responses to your questions and others.

    My approach is simple. I place high probability trades. I take winners early. I try to limit my losses. I make decisions based on daily moves as well as some guidance from the Greeks. I manage risk up to a 15% up move and a 20% down move so that I will survive to trade another day. I expect my strategy to have annual returns of 25% to 36% per year. As you can see I am underperforming my expectations at this point. There's no magic here at all. I know absolutely nothing!
    #86     May 28, 2016
  7. conduit


    You may want to multiply your daily pnl by 250 or so but not 365 as there are less trading days in a year. Also if you really want to obtain such projection you could perhaps extrapolate the cummulative pnl over the rest of the year as a daily projection will get you a very volatile figure that may not mean much at.

    From your sheet it is then still confusing what starting balance you kicked off from. Is it 20k or 150k or...

    In my comments re risk I assumed a balance of 20k.

    #87     May 28, 2016
  8. I'm basing my profit/loss over calendar days so I want to keep the formula as it is. The starting balance was $150k.
    #88     May 28, 2016
  9. Two trades today. I sold to open the 15 JUL 16220C(1)/193P(2) ratio strangle.

    One of my GTC orders triggered closing the 15 JUL 16 225C(5)/185P(4) for a 50% profit.

    The experiment is up $95.91 for the day.
    Theta is 122.52, delta is -66, and vega is -686.

    The experiment is up $2,965.71 for May (+1.97%).

    Here's a link to my spreadsheet.

    Happy trading!

    #89     May 31, 2016
  10. DTB2


    Take a look here:


    Can anyone backtest a touch of 1SD and then a touch of 1/2SD on the other side? Would be much fewer trades but seems like it would have ridiculous %
    #90     May 31, 2016