M.A. crossovers

Discussion in 'Strategy Development' started by Tom1234, Sep 21, 2007.

  1. Tom1234

    Tom1234

    If anyone uses the M.A. crossovers as an entry signal, what do you for confirmation?

    I'd like to find out the best confirmation indicator for Moving Average crossover signals.

    thanks
     
  2. timbo

    timbo

    There isn't any.
     
  3. See my answer on this thread:

    http://elitetrader.com/vb/showthread.php?threadid=104650
     
  4. Rocko1

    Rocko1

  5. Murray Ruggiero

    Murray Ruggiero ET Sponsor

    All seriousness, dual moving averages systems are totally unreliable. I know some long term traders who at least until five or six years ago used an 4 moving average crossover system with walk forward analysis. If you want to learn what walk forward analysis is, I have a tutorial you can download on my site.

    http://elitetrader.com/vb/showthread.php?s=&threadid=100883


    My research showed that three MA systems are a lot more stable than a dual moving average system. In addition triple moving average systems are almost as good as using four moving averages. I have written several articles showing the 3D optimization surface of both a dual average and triple moving average system to show they are not the same beast.
     
  6. MGJ

    MGJ

    Sure, lots of them. But there is a bit of a philosophical problem of definition: what's the difference between "entry signals" and "confirmations"? If you use Indicator X along with Moving Average crossovers in a trading system, perhaps something like this:
    Code:
    IF( (moving average crossover = TRUE) AND (Indicator X = Bullish) ) THEN enter long
    then the MA cross and Indicator X have the same function: each one is half of the (combined) entry strategy. You could just as easily say that Indicator X is the entry signal and MA Cross is the confirmation.

    In fact you could choose "Indicator X" to be a third moving average, and now your "MA Crossover with Confirmation" system is identical to Murray Ruggiero's "Triple MA" system.

    But if you are going to do some computer backtesting studies of (MA Crossover plus confirmation idea X), I suggest you start off trying confirmation ideas that don't include moving averages. For the sake of diversity if nothing else.

    Good old Chuck LeBeau suggests several ideas in his excellent book (link). See especially pages 85-101. LeBeau certainly agrees with your idea that it's vastly preferable to add another qualifying condition to the basic MA crossover signal. Whether you call it a "confirmation" or a "filter" or a "setup before the trigger" or a "qualifier" or a "part of the combination" is mere linguistic preference.

    Here are a few more, that you can program into your backtest software and try for yourself:
    1. (for longs) when the moving averages cross over, don't enter on the close. Instead, wait for price to break out above the high of the day the MA's crossed. Enter on a Stop order at price = (High of crossover day)
    2. Require the N-day ADX to be rising
    3. Require the N-day ADX reading to be above XXX value, and also require ADX to be rising
    4. (for longs) After the MA's cross, don't enter until you see N closes in a row, above the fast MA
    5. (for longs) Require the N-day momentum oscillator to be >0
    6. (for longs) Don't enter until the fastMA has risen for N days in a row
    7. (for longs) Don't enter unless the N-day RSI is above XXX value
    8. (for longs) Don't enter unless the close is very near to the highest high of the last N days. You can accomplish this using the Stochastic Fast %K: don't enter unless Fast %K is 85 or greater.
    9. (for longs) Don't enter unless the close is beyond the XXX-th Bollinger Band. You can accomplish this using Bollinger's Oscillator %B: don't enter unless %B is XXX or greater
    10. and so forth. Use your imagination! Don't be afraid to try out some ideas.
     
  7. GaryN

    GaryN

    Following this line of logic the more ma's you use the better.
     
  8. I remember reading about systems that change position size as price value crosses each moving average. I do not recall the system being more profitable, it was just one approach to scaling up.
     
  9. What are you really asking? If the fast moving average value is greater than the slow moving average value then the price value is likely increasing. That is all that it says. Nothing can predict the future consistently correctly and reliably. I think the idea is to have a position in every big price movement.

    If you are looking for an indicator that increases the percent of winning trades I can tell you how to do that. When price value shows a 5 % profit sell half the position and move the exit stop to breakeven. My tests show this method wins about 70 to 80 % of the time. Overall profit is about the same, maybe a little less because if a really big trender appears, well you sold half the position for 5 % profit so you don't make as much money. Plus there are more trades so the costs of slippage and commission are a little greater.
     
  10. Murray Ruggiero

    Murray Ruggiero ET Sponsor

    This is not true , the problem is a dual moving average system has problems due to phase relationships with the market. If you use a 1/2 cycle moving average minus a full cycle will be 180% out of phase with the market. Buying every top and selling every bottom. The triple MA does not have this phase problem so that why it works better.
     
    #10     Sep 22, 2007