Short Premium Research Dissection (Part 9)

As stated in the last paragraph of my previous post, I want to track the author’s order of optimization. First, she looks at DTE (30, 45, 60, and 75 DTE). Next, she looks at entry IV (VIX < 14, 14-17, 17-23, and > 23). Third, she looks at % of trades to hit 25% and 50% profit targets stratified by same IV (just mentioned). Fourth, she varies allocation (2.5%, 5%, and 10% of the portfolio with each trade) based on -100% loss limit and 60 DTE. Technically, she isn’t optimizing yet since she has yet to choose particular values for these four independent variables.

The next sub-section discusses the impact of changing the stop-loss (SL) level. She includes a graph and the following table, which I began discussing in Part 7:

My standard critique applies. I don’t know the exact methodology behind these numbers. I don’t know the exact date range (somewhere around 2007-2018 may be estimated from graph), but I do know sample size this time (second row). I don’t have enough statistics. I’d like to see things like distribution of DIT and/or losses including max/min/average [percentiles], average trade [percentiles], average win, PF (see below), max DD %, CAGR, CAGR/max DD %, standard deviation (SD) of winners, SD losers, SD returns, total return (which can only be estimated from the graph), PnL per day, etc.

I still don’t know what to do with median P/L as a percentage of premium collected (see Part 8, paragraphs 4-5).

I also don’t know what “median account return potential” is and she never mentions it in the text. Return on what? Buying power reduction (BPR) would make sense but she hasn’t mentioned or defined that (see final paragraph), either.

She says more contracts is more leverage, which can become more dangerous (“particularly overnight”). I’d like to see average BPR to better understand leverage. She says SL level is inversely proportional to number of contracts. She also claims number of contracts to be directly proportional to the frequency of SL being hit, but I don’t see win rate increasing from left to right along row #3. Hmm…

My eyes agree that the -50% SL equity curve (not shown here) is most volatile, but I’d like to see a supporting metric.

Finally, she claims larger SL levels allow for a better chance to exit trades closer to the SL—presumably cutting down on those excessive losses discussed in paragraphs 2-3 of Part 7 (linked above). She never quantifies the excess loss, though, which leaves its impact hard to understand. I can only shrug my shoulders at this.

I see a lot of mess and very little precision here.

Here’s another challenge to her reasoning about excess loss: who cares? Looking past the nebulous normalization details, a loss of 220% is 10 times the median P/L of 22%. Regardless of SL level, that is devastating. This suggests another metric to monitor: median PnL / worse PnL. As listed seven paragraphs above, I then want to know more about the distribution of these losses to see how often and in what magnitude they occur. This suggests median PnL / average loss, which hints at PF.

I will continue next time.