In our quest to understand trading options for income, the topic of this article is that conclusions based on inadequately designed research, presented by third parties, can often be more harmful then helpful. Recognizing some of the pitfalls can help you avoid costly assumptions based on faulty or inadequate research.
Let's review an example of a recent test conducted by a third party to better understand some of the issues.
A test was conducted for the purpose of comparing one strategy against another. When doing a comparison, it is important to keep all parameters the same and just change the strategy. It is also helpful to have a large sampling to keep the standard error low; typically a sampling in the hundreds, but no lower than thirty.
The strategies being compared were the directional IC vs. the skewed IC. Since the direction of the underlying is being considered, the two strategies were broken into a bearish and bullish bias.
The bearish directional IC (iron condor) is defined in the test as follows: a short Call spread where the short strike is located at the 70% POTM (Probability OTM); a short Put spread where the short strike is located 80% POTM. This presents a bearish bias because the short strike of the Call is closer to ATM (at the money); 70% vs. 80% POTM.
The bullish directional IC is defined in the test as follows: a short Call spread where the short strike is located at the 80% POTM; a short Put spread where the short strike is located 60% POTM. This presents a bullish bias because the short strike of the Put is closer to ATM (at the money); 60% vs. 80% POTM.
The bearish skewed IC is defined in the test as follows: locate the short strikes where you collect one-third the width, and on the Call side, widen the width of the spread an additional strike. The larger width on the Call credit spread increases the capital at risk making the IC bearish.
The bullish skewed IC is defined in the test as follows: locate the short strikes where you collect one-third the width, and on the Put side, widen the width of the spread an additional strike. The larger width on the Put credit spread increases the capital at risk making the IC bullish.
To determine if the trade will be bullish or bearish, the criteria was to enter a contrarian trade if within a 2-week period the underlying moved 5% or more. If the underlying moved up over 5%, then a bearish trade was placed; and vice versa. The underlying was SPY, and the data used was over a 5-year period. The trade was placed using a Monthly option with around 45 DTE (days till expiration). Positions were held to expiration.
The test yielded 21 bearish trades and 14 bullish trades.
Since we are focusing on the issues with the test, the results are not relevant.
So, what are the issues that would cause us to ignore the results?
First, the test introduced a second strategy: determining if the market is bullish or bearish and entering a contrarian trade accordingly. Since we have no data indicating how accurate (or inaccurate) this strategy is, we cannot determine the effects of this strategy on the results. A better approach would be to ignore market conditions and simply place bullish and bearish trades periodically (monthly).
Second, when testing the directional IC, the parameters for the short strike closest to ATM (and providing the bias) are different (60% vs. 70%). To properly compare the bearish and bullish results, they need to have the same parameters (either both at 60%, or both at 70%); otherwise the results will be skewed.
Third, the location of the short strikes for the skewed IC differ from the directional IC, resulting a lower POP (probability of profit). At one-third the width, the short location is actually below 1 SD (closer to one-half standard deviation). This makes it impossible to compare the two approaches since the POPs are different. To provide a fair comparison, the location of the short strikes between the two strategies should be the same.
And fourth, the total number of trades are too small: 21 for the bearish trades; 14 for the bullish trades. With such a small sampling, the results will have a large standard error calling into question the figures. If we ignore the requirement for identifying direction (5% move within a 2 week period), the sampling size would be significantly higher at 60 bearish (vs. 21) plus 60 bullish (vs. 14) trades (5 years x 12 monthly trades each); that's 120 total trades vs. 35 total trades.
In conclusion, as a result of the lack of controlling inputs and the small sampling size, the results of the test would be inconclusive. This underscores the need to properly design a test to ensure meaningful results.
If you would like to learn more about options, and how to generate consistent weekly income trading options, go to Options Annex.