Trading System #3–Naked Puts (Part 4)

In http://www.optionfanatic.com/2012/12/10/trading-system-3-naked-puts-part-3/, I showed how DD can limit position size and annual return despite a high profit factor.  The next thing I want to do is conduct a thorough DD analysis.  I want to see the distribution of DDs to see if the max DD is typical.  I then want to look closer and see if I might be able to minimize DDs.

Before I expend too much energy, I need to boost the sample size to enable statistically valid conclusions.  These 113 wins and 16 losses (http://www.optionfanatic.com/2012/12/06/trading-system-3-naked-puts-part-1/ ) compose just one study.  I generated every data point by taking account values at Friday’s close (Thursday in case of holiday).  I could rerun the backtest and take data points on any other day of the week, which would generate five studies.  I could also run Monte Carlo analyses to randomize the trade order.  With hundreds to thousands of Monte Carlo simulations, I can get distributions of performance statistics and study values that encompass 95% of the runs to feel more confident about my conclusions.

I must also flush out the arbitrary and make sure these results weren’t simply a matter of luck.  Rather than only backtesting a weekly stop-loss of $30K and staying out of the until two positive weeks are seen, I should backtest a range of values (e.g. weekly stop-loss of $20K-$40K in increments of $5K; remaining on the sidelines until one, two, or three consecutive weeks of market gains have been printed).  I could study the subjective function (RAR/MDD) and look for plateau regions rather than spike regions across the graphs.

Another thing I might consider aside from using a weekly stop-loss could be a cumulative stop-loss.  That is, if DD reaches a certain value then exit whether it takes one week, two weeks, or more to hit that DD.