Option Alpha's Post/Backtest on Earnings Trades

[FrankTheTank]

Was hoping @kim

[FrankTheTank]

Just now, FrankTheTank said:

Was hoping @kim

Sorry cant seem to edit my original post.   I was hoping @Kim would post a rebuttal or comments on this like he has done for some of the Tasty Trade earnings posts.  

https://optionalpha.com/we-stopped-trading-earnings-after-we-saw-this-new-research-188320.html

 

Thanks. 

 

 

[FrankTheTank]

[Kim]

We will let others to do their studies while we continue proving year after year that our strategies work.





https://steadyoptions.com/forums/forum/topic/5914-2019-year-end-performance-by-trade-type/

[FrankTheTank]

For sure.  I just wish he was able to break down his study into sub-sets of stocks such as lower IV vs. higher IV or by option volume.   Clearly the stocks SO uses are better than this raw average but the question is why?    Just trying to figure things out from a fundamental level.  

 

 

[Flameirinhas]

5 minutes ago, Kim said:

We will let others to do their studies while we continue proving year after year that our strategies work.





https://steadyoptions.com/forums/forum/topic/5914-2019-year-end-performance-by-trade-type/

It would be good to see the broker paper to check the real prices filled and the commissions and fees included.

[Kim]

5 minutes ago, Flameirinhas said:

It would be good to see the broker paper to check the real prices filled and the commissions and fees included.

The screenshot of real prices are included in every trade alert. Commissions differ between brokers.

[Kim]

11 minutes ago, FrankTheTank said:

For sure.  I just wish he was able to break down his study into sub-sets of stocks such as lower IV vs. higher IV or by option volume.   Clearly the stocks SO uses are better than this raw average but the question is why?    Just trying to figure things out from a fundamental level.  

 

 

Well, the big difference is that all those "studies" use the whole universe of stocks, enter at certain dates, use EOD prices and exit a day before earnings.

We on the other hand enter only certain stocks that show good results in backtesting, enter only when the price is good compared to previous cycles, and manage the trades actively by taking profits when our profit targets are hit. Makes a world of difference.

I tried to post a comment on the article, but after showing "pending" for few minutes, it disappeared.

[Arnold1321]

Yup, testing this kind of strategy with EOD quotes is meaningless.

[Options18]

I’ve actually done an analysis on all of Steady Options “calendar” trades dating back to the beginning, in 2011.

Came up with a 77% win rate and an average gain (including the losses) of 12.25%.

That is ridiculously good and almost too hard to believe..but it certainly is true.

I actually only trade calendars, straddles and strangles.   Nothing at all wrong with the other types of non-directional type trades..just my preference.

[Kim]

24 minutes ago, Options18 said:

I’ve actually done an analysis on all of Steady Options “calendar” trades dating back to the beginning, in 2011.

Came up with a 77% win rate and an average gain (including the losses) of 12.25%.

That is ridiculously good and almost too hard to believe..but it certainly is true.

I actually only trade calendars, straddles and strangles.   Nothing at all wrong with the other types of non-directional type trades..just my preference.

Some people just don't realize how unique is what we do in terms of returns and consistency. 

[Kim]

Here is a classic example how real trading is different from "studies".

On March 2 2:30pm we entered CPB straddle:



The price was 3.05 or 6.5% RV. We based our entry on the RV chart:



We exited the trade on March 3 10:05am for $3.45 credit, 13.1% gain



EOD price on March 2 was 3.40 and EOD price on March 3 was 2.95. The study using EOD prices would show 13.2% LOSS while our real trade was closed for 13.1% GAIN.

Two points that contributed to the difference:

1. Based on historical RV charts, we would not even be entering this trade at 3.40.
2. On the last day, we did not wait till the EOD and closed the trade in the morning.

This is just one example how a "study" can show dramatically different results from real trading.

[Kim]

On a related note, posting dollar P/L is meaningless - this alone disqualifies the whole study. The only thing that matter is percentage amount. Why? Because in order to get objective results, you need to apply the same dollar allocation to all trades.

Now, lets take an example of stocks like AMZN and GM. AMZN straddle can cost around $200 and GM straddle around $2. If AMZN straddle average return was -10% or -$20 and GM average return was +50% or $1, the average return should be reported as +20%. In the study, it would be reported as -$9.5.

[FrankTheTank]

On 4/25/2020 at 9:25 AM, Kim said:

On a related note, posting dollar P/L is meaningless - this alone disqualifies the whole study. The only thing that matter is percentage amount. Why? Because in order to get objective results, you need to apply the same dollar allocation to all trades.

Now, lets take an example of stocks like AMZN and GM. AMZN straddle can cost around $200 and GM straddle around $2. If AMZN straddle average return was -10% or -$20 and GM average return was +50% or $1, the average return should be reported as +20%. In the study, it would be reported as -$9.5.

Yes.  Tasty trade does this too and it drives me crazy when they include AMZN in their limited studies.   You would think their team of "data scientists" would catch this and they probably do but I suspect Tom tells them how he wants the results to look and then they scramble around trying to make it fit his narrative.  

 

[Kim]

50 minutes ago, FrankTheTank said:

Yes.  Tasty trade does this too and it drives me crazy when they include AMZN in their limited studies.   You would think their team of "data scientists" would catch this and they probably do but I suspect Tom tells them how he wants the results to look and then they scramble around trying to make it fit his narrative.  

 

Well, I didn't want to mention tastytrade as some people say I'm biased. But since you did.. 

I have no idea why they do what they do. Presenting results in dollar terms is completely meaningless. Even if you backtest the same stock over 5 years period, and the stock price doubled, which probably caused the typical straddle price to change significantly as well. So if 5 years ago the straddle price was $10 and today it's $20, how does it help to compare the dollar P/L? it's comparing apples to oranges. Even on the same stock.

[DubMcDub]

My basic thought on that article is that it didn't test SO's strategies or really anything all that close to them. Systematically and indiscriminately buying straddles on a predetermined universe of stocks X days before earnings and selling them Y day before earnings (or after earnings) is not what we do here.