Moving average on equity curve:

Discussion in 'Risk Management' started by amigasearch, Dec 16, 2003.

  1. Hi. Is anyone doing this? Can you relay your success / non success?
    It makes logical and perfect sense to calculate a simple (n) period MA on trades. If trades results are above the MA, then trade the system, otherwise, if below, wait until "simulated" trades turn proitable.
    I would love to hear developers experiences. I am considering this type of method.
  2. chessman

    chessman Guest

    I have toyed around with this idea as well. This would be similar to trend following, stay out during down downturn and ride the uptrend. A successful implementation would depend on the trends in the eq curve.

    One of the issues I thought about is what if the eq curve is just below the ma and a big trade happens, you miss out on it. This could have a substantial impact on overall performance. So I thought why not just reduce the size of the position during a downturn, this seems to work a little better than no position at all.

    The other headache, during the time the account is going sideways, you would tend to get lot of 'signals', position sizing based on signals could have an impact on performance. For example if the eq curve just crossed the ma and gives you a trade, you take it, this trade turns out to a loss. You skip the next trade because eq curve went below ma, however this trade is a profitable one. So the net result is we lost money rather than being even.

    I think the only place this strategy would do well is if a system has steady prolonged dd, then this would make some sense.

    Interesting topic, lets see what others have to add.
  3. Quah


    Isn't this sort of like saying "Go long. Sell when it goes higher. If it doesn't go higher, don't go long."?
  4. I like this idea, as I always wish I can see only the profitable signals! :confused: :D
  5. My quote:
    If trades results are above the MA, then trade the system, otherwise, if below, wait until "simulated" trades turn proitable.

    This is all under the assumption, that the MA of equity curve is lagging, which it always is.

    With that in mind, then Chess was right, many trades, if profitable will be lost when below the EQ curve.

    I figure I will know when to start buying once Equity line is above the MA.

    I have a backtested system, that has a straight eq line. BUT There are periods of DD, that will clearly trade below the MA.

    I will need to program this (not difficult) to see if my profits improve. I will let all know.

    It seems, that even with the lag, if I miss the "lower portion(the losses in a continued DD)" of this DD (due to the lag), then mission accomplished.

    That is good - Keeps me out of the rest of the DD - of course, I dont know how long this DD will last. Once simulated (trades system will watch and count not real profits on) trades cross above the MA, then start trading again.

    It seems in thoery to work. (I ask before I program to see if I can get out of programming it!). When MA crosses, then in theory, I am out of DD and am going to continue my smooth to the moon equity line.

    I was thinking of even implementing an amount below the MA (a percent amount) to stop trading, since whipsaws may occur.
  6. Quah


    Why use an MA?

    Why not a stochastics or CCI of your equity curve?
  7. This sounds good!
    I'm just thinking aloud. Maybe I will add an ErgoDic.
  8. I don't like to give a direct answer but here's one thing you guys have to think about.

    Are each trade independent to each other or not?

    Are there any under-lying relation or sequential significance between trades?

    If you can't figure that, the topic of using TA in general is useless.

    This doesn't apply only to equity curve but markets and TA you use.
  9. A trader can be only as good as his last trade!? :confused:
  10. ig0r


    Just compare current equity to the equity x periods ago, based on how much below or above, write a function to dynamically size positions, equity movements will turn somewhat parabolic
    #10     Dec 16, 2003