Adding exit rules dont work....

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

  1. Hi. Any thoughts on this:
    I have a system I am ready to trade. Takes 1 trade a day (avg. sometimes 2).
    Proitable (on paper of course!). Very comfortable with it.
    Takes trades from 9:30 - 3:30, and always exits 4 PM.
    1 minute data only.

    Now the question:
    This system preforms best only with initial stop loss (around a few hundred dollars), BUT NO profit taking rules. If I add profit taking rules, system still preforms well, but not as good (expectancy drops a few cents, which affects all others like profit factor).
    Now this is risky (in theory, no profit taking rules). I see many trades on my MFE plot (MFE plot that turn into losers - this year, for example, 2 11+ gains turning into less gains).
    But, system is ok on paper without any parameters.
    I have tried many profit taking rules, like adding indicators (adaptive MA, ADX, etc) but none are right for telling me to exit (usually, price always returns to where I was before).

    Would you trade this with just initial stop OR would you be on safe side to protect profits, by adding rules?
    Now, I am leaning toward just initial stop and thats it.
     
  2. dbphoenix

    dbphoenix

    Research previous posts on scaling out.
     
  3. acrary

    acrary

    What you're noticing is the normal behavior in the SP/ES markets.
    If you're on the right side of the market, exiting before the close results in sub-optimal performance because the high and low range for the day is normally completed very late in the day.
     
  4. My best stuff exits EOD on winners. LW has been a big proponent of using the close as an exit for winners as well.
     
  5. Thanks to all. I am going to try the scaling stuff today (first need to program this). I am "dreaming" the results will be positve, as it does make sense. All of my testing has been with one contract, so now I will need to add multiple (along with position sizing / money management rules).
    Acrary, your analysis proves a very substantial point, and enforces my belief in this system. Thanks again.

    Being new to system dev. the one thing that has suprised me the most is lack of TA indicators to use. I thought for sure they would help as filters of some kind for exits, but no success with these. I have just tossed them out (my app. will run quicker without the extra indicators running in the background). Your posting, Acrary also reinforced this concept.
     
  6. dbphoenix

    dbphoenix

    If you'd rather back away from indicators, you might also look at support and resistance and how volume behaves at these levels. Friday, for example, the NQ halted at support. If scaling in and out, this would have been a logical place to exit one contract. If one had simply left things alone, of course, he would have ended up at the end of the day more or less where he began, making the entire day a waste of time.

    There were also ops to short in the top of the base and go long at the bottom before the afternoon runup, but that's another subject. The point is that scaling can enable you to take advantage of ops that you might otherwise have to let pass by.
     
  7. Again, thanks for the insightful post.
    As I said, i am in the "optimizing stage". So far, no optimiztions have worked - this system works ok on its own (no indicators to support it, trailing vals at different profit levels, price moves, etc).
    The support / resistance idea I will also try (every time you post DB, you are increasing the amount of coding I need to do!). It is a new idea I have not thought of.
     
  8. dbphoenix

    dbphoenix

    Sorry, I didn't realize you were coding something into that sort of system. Bringing S/R into it probably wouldn't work because what I described would involve trading other traders' emotions, not a mechanical system.
     
  9. Yes. App is C++ - dont worry, I dont mind the work (as long as system pays off).
    About Support / resistance levels, and Mechanical. My thoughts on this - I can probably code in points over the last n minutes / hours, days, whatever, and gauge throughout the day where the support and resistance are, based on prices trading there. A sort of indicator to tell me where followthrough was not allowed (resistance) or bouncing off (support).
    I am a strong believer that, to a degree (and this is a discussion that can go on forever), given the rules, it can be coded. Its just the amount of code that is the problem. When I mention to a degree, though, yes, there are certain points on a chart I cannot express in code, but so far none of this lack of expression has hurt my moving ahead.
     
  10. ...don't sweat it that your BT says you shouldn't set a profit target. Although I don't have any opening hour NQ systems like that, the profit targets on some of them are so wide (20 NQ points) that they might as well not be there. That just means it's a really good system.
     
    #10     Dec 14, 2003