Short Premium Research Dissection (Part 17)

Last time, I concluded critique of our author’s unlimited-risk strategy.

The next section begins:

     > The short iron butterfly and short iron condor are limited-loss
     > strategies, but can be highly risky when using large trade size.

I like this caveat. For many of us, an unanswered question lies between “unlimited risk” and “limited loss.” That answer has to do with position sizing, which is completely individual.

     > None of the content below is a recommendation to implement an
     > investment strategy. Rather, the research below is meant to help
     > you make more informed trading decisions, and learn systematic

I concur with this disclaimer.

     > trading strategies with historically-favorable performance.

Does she mean to imply that she tweaked the data to make the performance look favorable? Hopefully she never had any intent to curve-fit (even though I think she did as discussed in Parts 13–14) despite the results turning out to be profitable.

In describing the next strategy, she writes:

     > We’ll… [construct] with a long 16-delta put, short ATM
     > straddle, and a long 25-delta call. As we’ve been doing,
     > we’ll look at options closest to 60 DTE.

The setup includes a fixed-delta legs, which I think is suspect. This is an asymmetrical butterfly with an embedded PCS. My preference would be to study various permutations rather than just the 16-delta put (see second and third paragraphs here). Assuming she has done this honestly—by writing up the research plan before looking at the results—this is not curve-fitting. It does leave the door open to fluke, however, in case these particular trade parameters perform well while adjacent ones do not.

As I have written about extensively (e.g. here and here), trust should always be an issue in the financial industry for reasons including widespread precedent. You may, but I will not (see second prerequisite here) assume she has been totally honest. Charging money for this research is an underlying motive to present positive performance. I want to see a strategy that performs, in large sample size, with variable-delta legs. I don’t need to see everything do well, but I should be able to discern a range of encouraging parameter values. As described in the first paragraph here, I want to see something honest.

The author explains two potential sizing methods:

     > …based on the actual maximum loss is much safer because
     > as long as you size the position correctly, you can’t suffer
     > a loss larger than you’re comfortable with. When sizing…

Maximum loss would be incurred when the wider credit spread is ITM at expiration.

     > based on a certain percentage loss on the premium received,
     > it’s very possible to incur larger losses than you wanted
     > because nothing guarantees trades can be exited exactly at
     > the stop-loss level.

This is the “excess loss” that was last discussed below the first excerpt here.

Contract size will be similar either way when max loss approximates credit received (i.e. max profit). I would like to see a histogram of max profit : max loss. If a significant difference exists most of the time but max loss is extremely rare, then it may be worth position sizing based on max profit (and limited allocation) despite the increased risk.

I will continue next time.