Can Price Action Trading Be Automated?

Discussion in 'Automated Trading' started by SimpleMeLike, Dec 26, 2016.

  1. I often hear people say that price movement can be programmed for analysis. Not being able to do so myself, I can't disagree. However, the quality of examples of such that I've seen is still light years behind that which can be done manually. And given the need for analysis of context, and the fact that price movement is relative, there is a certain amount of judgement and experience required, which I'm told computers do not yet have. Attached are examples I found of the automated drawing of something as simple as a trend line from eSignal and Tradestation etc. Not very impressive. What a program would do with something complicated is hard to imagine. Maybe there are 'top secret' TA computer programs that can outperform us, I don't know. But until available, manual screening/analysis/backtesing still seems best to me.
     
    • tl1.JPG
      tl1.JPG
      File size:
      76.7 KB
      Views:
      142
    • tl3.JPG
      tl3.JPG
      File size:
      35.3 KB
      Views:
      137
    • tl4.JPG
      tl4.JPG
      File size:
      56.5 KB
      Views:
      134
    Last edited: Dec 27, 2016
    #21     Dec 27, 2016
    SimpleMeLike likes this.
  2. Handle123

    Handle123

    Define trend without moving averages so you have one less problem of coding, base entries using tick data instead of one minute bars as there are times when to enter and get stopped out on same bar, you can't tell if entry was done first or second using one minute bars., if you using futures, I don't recall if Ninja allowed you to automatically switch contract months which can lead to other problems dealing with indicators not being able to rely on cause lack of volume. Whatever you can dream up can be programmed so long as you can think of the ways to program. Ninja has a good support group and some outstanding other traders who know how to program and might help when you get stuck. I recall Ninja has several things it would not do and why I left them, at the time you could only have 500 symbols to bring in data and using the screener would really slow down the software and freeze, so if you want to check 10,000 stocks I couldn't and if you wanted to see mismatching of options, forget that. And had to put in symbols by hand. But for most things, they are good.
     
    #23     Dec 27, 2016
    SimpleMeLike likes this.
  3. comagnum

    comagnum

    If you really study the charts paying attention to what the institutions are looking at than you can find repeating areas where institutional buy side programs kick in and send the price trending up over days to weeks. It is subject to change over time so you will have to adapt over time.
     
    #24     Dec 27, 2016
  4. There is nothing mysterious in the charts that you show. In these charts, I see trendlines through recent lows and highs. This is elementary to code and to automate. Far more complex price and behavioral patterns have been quantified, coded, and automated.
     
    #25     Dec 27, 2016
    SimpleMeLike likes this.
  5. But the fact of the matter is that the programs drew so many of the lines incorrectly that using such for screening, analysis, back testing or trading would be a disaster. And as you said, if the trend line is elementary to code, what would the results be when judgement is required in addition to being more complicated than simply connecting two points correctly.
     
    #26     Dec 27, 2016
  6. When you say "judgement", you seem to be implying that it's something so "human" that it can't be coded. Sure, some people may have a "gut feel" (i.e. the intuition) that makes them successful traders, and that would indeed be very difficult to quantify. But setting that aside, much of what's called "judgement" can be coded.
     
    #27     Dec 27, 2016
  7. Hello learner2007 and thanks for the comments. Very logical.

    Currently, I trade similar to you: discretionary screening/analyze and lots of time manual backtesting an idea. I agree with you, but there is some limitations on my part with manual backtesting.

    Here is the problem I (maybe others) have with backtesting manually: Although good experience cause I get alot of charting practicing on price action (trendlines, channels, support and resistance, just seeing the price in action, etc) and ideas pop in my head during backtesting manual. The ultimate problem is time and incorrect backtesting.

    For example, lets say I am forward or backtesting an idea strategy day for day. A month from now, I may realize the strategy needs another testing parameter. Then i have to start all over again manually backtesting.

    Currently, I have been forward testing an idea I got stuck on for 4 months, all to realize like Handle123 stated above, I have no way to define the trend. I just look a price direction with the eyes and proceed with entry/exit/money management rules. When I first started, i was thinking of discretionary trading the plan. Well guess what? Would it not be better/efficient if I could just program the strategy and back test it on a platform in a matter of hours to see if idea profitable data?

    Do you manual backtest your ideas?
     
    #28     Dec 27, 2016
  8. Hello Handle123 and thanks for the comments.

    This is a information to know as I use open and close candles to define entry.

    I trade futures only and still reviewing platform I will decide to use.

    Thanks,
     
    Last edited: Dec 27, 2016
    #29     Dec 27, 2016
  9. You might want to spend 100% of your efforts solving this. And once you do, I think you will find that 99.9% of the problems you're having will vanish on the spot.
     
    #30     Dec 27, 2016
    dartmus, KDASFTG and SimpleMeLike like this.