Adding exit rules dont work....

Discussion in 'Strategy Development' started by amigasearch, Dec 14, 2003.

  1. Hypo, your posts continue to aid my confidence.
    BTW - remember the advice you posted to me, maybe 1 month ago, the one where i let out some parameters? This is the evolution of it. Completely different, in style, but based on that one. Before I trade it, I will be sure to PM you.
    #11     Dec 14, 2003
  2. ...the reason I recommend weekly backtesting is that it reveals how the parameters are shifting with volatility increases, which according to my testing we are experiencing now. I will look forward with great anticipation to learning how your system development goes. If you read Schwager's Wizards books, you may recall that in one interview (I forget which) the trader said something to the effect that "There are about 20 simple but highly profitable patterns out there that if they were generally known, their edge would vanish instantly. Look for them. You'll find them."

    That's why you keep your mouth shut when you find something good. Of the systems I currently trade, one is so good I can't believe it, and it's also trivially simple. Just because you're a po' boy and an amateur doesn't mean that you can't find something big if you keep at it. Then one way or another you can scrape up the cash to trade it to its limits. IMO a really good system can grow a small account at 200-400% a year. Of course there are all sorts of personal issues that you have to deal with along the way, and portfolio-management-like issues to learn how to handle. Good fortune to you.
    #12     Dec 14, 2003
  3. Try to add exit rule based on average trading range, at20% of ATR exit one contract, at 30 % of ATR exit second one or any variation of it. If you like it, let me know how will this compare to your original system .
    #13     Dec 15, 2003
  4. jaan


    that's an interesting observation, but would you expect to see different numbers in your table if the underlying series followed a random walk?

    - jaan
    #14     Dec 16, 2003
  5. Programmed in "scaling out" techniques using multiple contracts.
    DB, good advice. Seems to work well for me. Thanks.
    #15     Dec 17, 2003
  6. They



    When using a systematic approach be mindful when incorporating "scaling out". Each scale out has to be treated as a separate system, especially when they all have the same entrance and initial risk. Of course you can compare them and maybe one of them actually does provide a better exit than MOC.

    Viewing them as separate systems and trading a basket of them can also be a good approach but is dependent on your resources.

    Otherwise don't forget your original post/system - no exit rule seems to improve the system.

    As far as ideas for exits go;

    2 or 3 std. dev. from mean of the day
    Volume spikes
    Premium spikes
    % Increase of trades per minute
    20 day high and low volume levels (where the most/least volume occurred in the last 20 days)
    ...Initiating a trailing stop after one of the above occurs

    Trade on and be excellent to each other
    #16     Dec 17, 2003
  7. acrary


    Here's the same test done with random SP data over the same time period.

    I see some big differences...maybe a opportunity to exploit. Need to check the stats to see. Thanks for the idea.
    #17     Dec 17, 2003
  8. Do you have a separate numbers when high of the day and low of the day are completed ? My guess is that highest% of high's or low's is in by 2:30
    #18     Dec 18, 2003
  9. Good luck with the idea! Pm me with your findings, if you can (out of curiousity).
    #19     Dec 18, 2003
  10. jaan


    you do? hmm.. the corresponding columns seem rather similar to me (after discounting the variance).

    of course, the proper way to analyse that would be to test the stat hypothesis that your real SP table is different from those of a random series of the same volatility (well, actually, my intuition says volatility is not too important here).

    and of course i'm too lazy to do that myself.. :)

    anyway, the point i wanted to get to was that *if* your SP numbers are not statistically different from those of a random walk, then they would *not* support your conclusion (it pays to hold the position until the close), because random walk (by definition!) ensures zero expectancy regardless of trader's actions.

    - jaan
    #20     Dec 18, 2003