Programming Breakouts

Discussion in 'Automated Trading' started by dunitlongpole, Apr 22, 2006.

  1. Hello.
    I am wondering if it is possible to automate a system based on Triangle, Channel and Momentum Breakouts? I can look at a chart and pick them out, but can I make a program to identify them and automate a trade?
  2. How long before the breakout do you want an incoming signal to you?
  3. When the breakout is occuring...
  4. The easiest combo is an MACD of volume and RSI.

    When you put it up on your chart you will see how the signal (spike) will give you a trigger.

    This is an indirect method of measuring when A/D goes from A to D.

    Personnally, I like a signal that is well before the BO and a more important signal is the exit trigger on the three formations you are checking out.

    If you are into IT position trading you could use the 20 day entry of the Turtles or you could pep it up with the rally attempt analysis of WJO of IBD type trading. The IBD test is a 4 to 10 day test that really aces the turtle 20 day test on both performance and R/R bases.

    Formations, unfortunately, by their nature are better for back end stuff than front end stuff. Thus you can see why I am giving your detectors that are precursors to the actual formations. I am not trying to dissuade you from your labors, just pointing out that there are a lot of easy ways to make a lot of money.

    The only way it is ever possible to get a person to move forward on a path is to be where the person is and work stuff out from there.

    My response covers the bases from rapid trading (5 min) all the way to intermediate term.
  5. toe


    If you can come up with a definate, unwavering set of rules for entries/exits then you can automate it.

    Are you actually trading this successfully already as a discretionary system, or is it an idea you'd like to test? You say you can see these trades on a chart, so I'll presume its the latter. This is a very good case for regression testing which is a precurser to automating. Create the rules and backtest them.

    Remember that just because you can see good trades with a set of rules doesn't mean there aren't plenty of bad trades with the same rules, if you're thinking about regression testing to see if your ideas are profitable and stable then your on the right track.

    But remember to learn all you can about avoiding curve fitting first. Curve fitting is like a broken clock that tells the correct time twice a day. Its easy to come up with a set of rules that will be profitable over some period of time, where in fact those rules made good trades by 'accident'.

    Here's a link to a channel breakout system in Wealth-Lab, you can run it on the web sight on any symbol, try AA, then check out some of the other systems to get an idea whats possible, but remember many systems are curve-fit.
  6. There are also a lot of people who talk out of their ass giving the impression all you need is a big bucket to scoop up the dough on wall st. yeah, it's really that simple.

  7. quote from grob109 here at ET on another thread recently

    snip... People will definitely lose money using my stuff. That is a fact for everyone to know. snip....
  8. I wouldn't recommend automatically trading such pattern matches - perhaps you are better off being alerted to a "potential pattern". The reason I say this is, I have been looking at automated ways to identify patterns in stock charts. Here you can see automated identification of "Head and Shoulders" patterns on daily data for stocks ->

    For that, I am just using simple correlation - so it certaily can be done - the question is - are the outcomes of the patterns tradable?

  9. nealvan


    I just hold out my staw bucket every day and people come by and fill it to the top... I thought everybody had it that easy...
  10. Hi, just wondering if there were other software available that could recognize patterns than the ones mentioned above. I'm currently using CyberTrader and IB as my brokers.

    I am currently profitable trading my system but want to trade only the best setups. Right now most of my trades are about 1:1 risk/reward ratio with about 70% success rate. A few times a week I get a great setup like 1:5 risk/reward ratio where I can do some heavy shares. So I would like a way of being presented with only these great setups all day long. My custom candle patterns have exact entry and stop prices that I can program in.

    I'm looking for a customizable application that would allow me to look for custom candle patterns as well as "preset" patterns (head&shoulders, flags, triangles, etc). I find that my custom patterns signal the breakout from the "preset" patterns and turn into a profitable trade, at least a little over 70% rate.

    Right now I can only manually track like 3-5 stocks closely enough during the day to trade them. What I'm looking to do is scan a few hundred to thousand stocks and then generate in real time several lists that alert me when:

    1. There is a matching formation with trigger prices listed (within .10-.15 of triggering)
    2. Formations that have actually triggered.

    This way I would be able to be presented with lots of good setups all day instead of looking at my 3-5 stocks and trying to squeeze less optimal trades from them.

    I think Tradestation can do this but I want to avoid them if possible since the commisions at IB and Cybertrader are lower. Plus I love the charts (and speed) of CyberTrader and do my futures trading with IB.

    If I can't find an off the shelf program then I will write a custom program myself using (Java or a .Net language) with IB's API. But I want to avoid this if I can to save time and energy.

    Is there a scanner or automated trading program out there that can do this? Is Tradestation the only one? How about Ninja Trader?

    #10     Jun 18, 2006