Price algorithm entries? (entering as a function of price instead of with TA or S/R)

Discussion in 'Strategy Development' started by 1a2b3cppp, May 16, 2013.

  1. What if instead of trying to identify support and resistance, you just actually tried to trade the noise.

    Price entry would be based on a position relative to the last tick of the last bar. Usually when price is going somewhere, there is still some noise because it doesn't go there in a straight line.

    Look at something like this. If you were to buy every few tick retracements from the high of the previous bar, as price made an uptrend you would accumulate a large position, most of which would go in your favor.

    [​IMG]

    So this seems initially like it requires knowing when price is going to go up for a while, which is prediction, and if you could do that you would just go long with a huge position in the correct direction. And of course, if price goes down then it doesn't work.

    I strongly dislike adding to winning positions because it raises your average cost and results in price only needing to make a smaller move against you to turn your winner into a loser, but adding to winners seems paramount in a strategy like this.

    And I suppose knowing when to exit is important, too.

    Let's discuss ways to be able to do something like this without needing to predict direction.
     
  2. dom993

    dom993

    By definition, the mean value of noise is 0, so the practical way of trading noise is by fading moves at the outer band of that noise.

    Defining price-action as signal + noise, in a trend (signal) entering in a pullback is a form of entry taking advantage of the noise (pullback). The ideal form of it : a channel.

    In a trading-range (no signal), fading the extremes is a form of entry taking advantage of the noise.

    In both cases, the key is detecting early enough a change of signal.
     
  3. by whose defenition?
     
  4. And then some
     
  5. eurusdzn

    eurusdzn

    If you traded noise only in the direction of the mean with equal profit stops and profit
    targets would not this be profitable IF, AND as long as market selection ( trend identification) is correct? Regardless of entry criteria.?
     
  6. technically yes..random entries with the trend should work
     
  7. And there in lies the fundamental problem with technical analysis. What defines a trend? When does a trend end, and when does it begin? Is a pullback a pullback or the start of a trend reversal?Trading with the trend works, until it doesn't. When it works, and when it doesn't are random events in time.

    Also, any entry method should be thought of as part of a trading system. A backtest of the trading system will reveal if it has positive expectancy
     
  8. I agree with all of that.
     
  9. Maybe you can get context to the noise with two period Bollinger Bands or something. Of course there are those times when price just goes straight up or down without retracing at all, so it's probably going to have the same issues as trading Bollinger Bands with a longer period. See this thread for a discussion about Bollinger Bands.
     
  10. Fix this by doing what all market masters have said to do. Start with monthly, then weekly, then daily, then to your choice of intraday time frames.

    Do not trade against the trend. WD Gann onwards has always said this.
     
    #10     May 20, 2013