Short Premium Research Dissection (Part 35)

Finishing up the sub-section from Part 34, our author writes:

     > At the very least, when incorporating short premium options
     > strategies… it may be wise to implement time-based exits
     > to avoid large/complete losses and reduce portfolio volatility.

By not stating definitive criteria, she [again] tells us nothing conclusive. As discussed in the second paragraph after the first excerpt here, we’ll have to wait to see if a time stop makes the final cut.

Unlike the 50% profit trigger for rolling puts to 16 delta, I think the time stop is beneficial. I wish she had shown us the top 3 drawdowns as done previously (e.g. Part 31 table). Seeing improvement on each of the top 3 would be stronger support for time stops than improvement on just the worst. Even better would be the entire standard battery (second paragraph).

Unfortunately, we do not know how time stops would impact the high-risk strategy, which was backtested before 2018.

I can’t help but wonder why time stops were not studied before 2018 (final excerpt Part 31). Admittedly, I would have suspected curve fitting had she written “due to the horrific drawdown suffered in Feb 2018, I tried to add some conditions to make the strategy more sustainable.” Not giving any explanation makes me suspect the same. I think the best approach to avoid being influenced by disappointing results is to plan the complete research strategy in advance (second paragraph below excerpt here). Exploring the surrounding parameter space (see second paragraph here) is critical as well.

     > The next trade management tactic we’re going to explore
     > is the idea of ‘delta-based’ exits.

Like time stops, this is another completely new idea for her. As just discussed, this ad hoc style bolsters my suspicion (see third paragraph below excerpt in Part 30) that she spontaneously tosses out ideas with the hope of finding one that works. Given enough tests, success by chance alone is inevitable. “Making it up as we go along” to cover for inadequate performance is a flawed approach to system development.

     > To verify the validity of using delta-based exits, let’s look
     > at some backtests to compare holding to expiration, adding
     > a time-based exit, and then adding a delta-based exit.

Layering conditions is confusing because order may matter* and all permutations are not explored (see second and third paragraphs here). For example, she does not study a delta-based exit without a time-based exit. I would rather see each condition tested independently with those meeting a predetermined performance requirement making the final cut.

     > First, let’s compare… holding to expiration… [with]…
     > exiting after 30 DIT.

Where did 30 come from? The exploratory studies (Part 34) looked at 10, 20, and 40 DIT.

When something changes without explanation, the critical analyst should ask why. This is becoming a recurring theme (as mentioned hear the end of Parts 32 and 33).

I will continue next time.

* In statistics, this is called interaction.