2024 - Modified Wheel Strategy using AI Signals

Discussion in 'Journals' started by frostengine, Dec 29, 2023.

  1. As part of a public experiment, I will trade a modified version of the Wheel Strategy and document the results here.

    Strategy parameters:
    - All entries are based on signals generated at https://unlockingalpha.com/

    - Weekly options will be sold, typically between .20 and .35 delta

    - If a cash secured put is set to expire in the money, the put will be rolled to the following week or assignment taken ONLY if unlockingalpha.com generates a buy signal for that day. Otherwise, the position will be closed out for a loss. This is different than the traditional Wheel Strategy which always takes assignment or rolls.

    - Unlike the traditional Wheel Strategy, cash secured calls can also be sold without owning the underlying stock when a short signal is generated. The same rules apply for rolling or taking assignment. If a signal is generated that day only.

    - If an opposite signal is generated, any short options against that signal will be closed. Example, if short puts and a short signal is generated, those short puts will be closed immediately.

    - If an option captures 80% of its premium, it will be closed early

    - The goal is to sell $1,000 worth of premium every day. Using the trading signals generated at 9:45, 11:00, 11:45, and 12:45 (all times central time). Due to long term market upward bias, selling puts will be preferred to selling calls when all else is equal.

    About https://unlockingalpha.com/
    - All signals are generated using AI and posted in real time to that site
    - Backtesting results from 2004 through June of 2023 are available
    - The system has been walk forward testing from June 2023 until December 2023
    - I started trading the system live 2 days ago using the above strategy. Below is a screenshot of the current holdings.

    For reference, in December it generated 1239 signals with a winning percentage of 75.95% and an average return of 1.57% per trade. This is in-line with historical norms.

    So far, I have sold $2,428 worth of premium the past 2 days. $179 was closed today on a short IWM call position that captured 80% of the premium.

    From the screenshot, you can see that $486 of the premium is currently already unrealized.
     
    Slow Learning Elf and Quanto like this.
  2. Why go through all the hassle? Just take 100% of the premium by taking assignment and hedge it.
     
  3. Quanto

    Quanto

    Keeping 100%? Is that possible at all when get assigned?
    Just prove your claim with a demonstration.
    You know: Extraordinary claims require extraordinary proof.
     
    Last edited: Dec 30, 2023
  4. Quanto

    Quanto

    @frostengine, how much was its average monthly PnL% in the said test phase(s)?
    Indeed an interesting experiment. Wish you and your this pet project all the best.

    Update: just found the answer at the given link: PnL 21.5% in 7 months means about 2.82% per month, and about 39.63% per year.
    Hmm. I would have expected much more, especially from such an AI :)

    https://unlockingalpha.com/
    unlockingalpha_forward-test.png
     
    Last edited: Dec 30, 2023
  5. TrAndy2022

    TrAndy2022

    Better stay away. I find 1.6% net profit per trade on options by far too low. It is all about execution and good fills. Otherwise that kind of profit could easily be vanished. You also need to consider that OOS performance is always lower so that 1.6% could be just 0.6% per trade than you should get cold feet or you know exactly how to get good fills everytime. Because on a bad trade you always get your fill because it is a loss. But on a good trade it could just get away witout getting filled. That would harm a lot your overall performance then. You really need to check bid ask spreads how wide they are and how you could get executed with minimal slippage. And do not believe their OOS showed performance it could be just optimized as often that is like in-sample. You need to track on your own OOS or with realtime tracking to get more true results here. Because if you just believe what they are saying you are just another victim of their marketing campaign, which purpose is selling it to potential customers and not making money from trading. My two cents.
     
    Last edited: Dec 30, 2023
    Quanto likes this.
  6. BKR88

    BKR88

    From the guy who claims to be an options expert. o_O
     
  7. Quanto

    Quanto

    You are taking the sentence out of context.
    What does doing so tell about you?...
     
  8. The 21.5% return is if you BOUGHT and held the underlining for 7 days when the signal was generated. Instead of trading the underlining outright, I am choosing to use the generated signals as part of modified Wheel Strategy.
     
  9. The 1.6% is if you used the generated signals to buy the shares. I am choosing to use the generated signals to sell options instead.
     
  10. destriero

    destriero

    lol how do you cash secure a naked short call? This may be the dumbest sht on the internet. Applying spurious “AI” with fantasy backtests to an inferior strategy. Just brilliant.

    So you plan on covering on LTD if the website doesn’t recommend the shares on LTD (or) isn’t currently recommending? You’re not clear on that point.
     
    #10     Dec 30, 2023