Thursday, July 31, 2014

Iron Condor Backtest - NDX - 38 DTE

In this post we will look at the automated backtesting results for four variations of a 38 days-to-expiration (DTE) NDX "no touch" iron condor (IC).  The backtests results for the NDX "no touch" ICs are worse than than those for the SPX and RUT "no touch" ICs, but I am posting these results for the sake of completeness. The prior NDX "no touch" IC backtest results posts can be found at:
As with the prior tests on the RUT and SPX, the NDX short strikes for both the call credit spreads and put credit spreads will be at approximately the same delta.  These backtests are "no touch" tests, meaning there are no adjustments during the trades.  In addition, there is not any hedging to start the positions leaning one direction or another.  Many of these settings are available in the backtester, but we will only look at this baseline IC cases initially.  See my post Thoughts on Options Strategy Backtests for additional background on my testing approach.  The setup details for this series of backtests are shown below:

(1) The backtester will start looking for trades that meet the entry DTE requirement after this date
(2) The backtester will not take any trades that will have an exit DTE after this date
(3) Some trading platforms call 38 DTE, 36 DTE (e.g., TOS).  The OSB uses a DTE based on the number of days to the expiration date in the option code/opra code, which is a Saturday for indexes
(4) Some trading platforms call 8 DTE, 6 DTE (e.g., TOS)
(5) Nasdaq 100 Index options
(6) Four 38 DTE "no-touch" iron condors will be tested with their short strikes at varying deltas (8, 12, 16, and 20)
(7) The distance between the short call and long call (also, the distance between the short and long puts).  This number was 20 for the RUT, 25 for the SPX, and is 25 for the NDX due to options availability.

The equity curves for each of the four delta variations are shown in the graph below.  In addition to the equity curves, the ATM IV at trade initiation is also plotted (the average of the ATM call IV and ATM put IV).  Please note that the dates in the chart are expiration dates, not trade initiation or closing dates.  For example, the trade corresponding to the 12/17/2011 expiration was initiated on 11/09/2011 and closed on 12/09/2011.  In all of the charts below, I will show data for trades by their expiration date.

If I pick a random expiration date from the chart below, for example 12/17/2011, we can see that when the trade was initiated, the ATM IV was 30.  When the trade was closed, the cumulative non-compounded profit had grown to between $13.4k and $32.5k depending on the short delta of the variation of the strategy.

For these trades, the maximum reg-t margin requirement is between $19.1k and $23.3k assuming your brokerage only margins one side of your IC.  If you have portfolio margin, your requirement is somewhere around one-quarter to one-half the reg-t margin numbers.  The 8 delta trade had the highest margin requirement, while the 20 delta trade had the lowest margin requirement.  The margin difference is related to the difference in the size of the credits received.  Higher delta equals larger credit; lower delta equals smaller credit.

In the table below are the standard trade metrics for the four ICs with different short strikes (8 delta, 12 delta, 16 delta, and 20 delta).

As with prior tests, the number of winning trades increases as the size of the short deltas decreases...the further away from ATM, the higher the win rate.  The combination of shorter DTE and closeness of the short strikes makes the higher delta condors have a lower win rate.  The 8 delta variation has the strongest summary statistics.

In the heat maps, we can see the performance by expiration month of each of the individual trades of each of the four delta variations.  The 0% cells represent expiration months were no trade was initiated.  Some of these 0% cells are due to lack of data or bad prints on the trade entry day...leading the backtester to skip that month for testing.  You can see the win rate decreasing as the heat maps move from lower delta to higher delta short strikes...more red cells.

We can also look at the 8 delta strategy in terms of the P&L range (in %), range of the underlying (in %), and IV range (in %), for each trade by expiration date.  This data is shown in the graphs below.  These graphs are showing the open, which is the 0% level, the high (green bar), the low (red bar), and the value at trade close (the blue line).  Also, note that the data in these charts corresponds to expiration dates, but the horizontal axis is not displaying expiration dates in order to make the charts less cluttered with labels.

The top graph displays P&L range for the 8 delta variation of the 38 DTE "no touch" IC.  Even with the short duration of these trades, you can see that when a trade closes profitably the range is mostly on the positive side of the graph.  With losing trades being exactly the opposite.

The middle graph shows the range of the underlying, the NDX, in percent terms during the life of each trade.  The bullish bias of the US markets is evident even in this short duration trade.

The bottom graph shows the ATM IV range and closing value in percent terms, during the life of each IC trade.  IV decays to zero as we approach expiration, so you would expect these bars to be mostly red, with the closing value to be negative (the blue line).  There were several times where ATM IV finished higher, and green...usually when there were market drops.

The next three graphs show the P&L range for the 12 delta, 16 delta, and 20 delta variations of this 38 DTE "no touch" IC strategy.  You can see the increase in P&L volatility as the delta increase from 12 to 20.

In the next post we will review the 52 DTE "no touch" NDX IC.  Drop me a note if you'd like to see different data presented, and I'll do my best to incorporate your suggestions in upcoming posts.

Also, if you don't want to miss new posts you can follow my blog either by email or RSS...both are free...and are available on the top of the right hand navigation column under the headings "Follow By Email" and "Subscribe To" respectively.


Vivek said...

Hi Dave, I have been following these backtesting results since some time. Thanks a lot for sharing this wonderful data with us. I am trying to test some ICs on Indian markets. Had just one query. Are these nos ex of comms and if yes then what assumptions are you taking for the same. Thanks again.

regards, Vivek

Dave R. said...

Hi Vivek,

Thanks for your question.

Yes, these results do not include commissions. Additionally, the prices for all strikes are the mid of the bid/ask spread.

My backtests attempt to compare the relative performance of strategies rather than trying to simulate "real" trading performance. In the latter case you would absolutely need to include commissions and have in place an algorithm to simulate slippage.

In my strategy comparison testing, I am assuming that slippage and commissions would be similar across all of the strategies that I am testing, and therefore I am removing them from the equation. Take a look at my blog post referenced at the top of this post for more information:

Hope this helps.


Post a Comment