Automated pattern recognition

Discussion in 'Automated Trading' started by ivob, May 20, 2006.

  1. ivob

    ivob

    Hello,

    Does anyone know of a software program that automatically detects certain patterns in price development such as W formations, cups, etc?

    Has anyone of you ever tried to write something like that and what are your findings? Do you think it is possible or would the human eye always be better in recognizing?

    I am a programmer myself (perl) and am considering starting a project but would like to find out what has already been done and what your thoughts are about this.

    Regards,
    Ivo
     
  2. lastick

    lastick

    Don't waste your time, it doesn't work.
    Nevertheless if you want to try it, choose a faster language than perl.
     
  3. Some textbook patterns including the ones you mention are available for Tradestation, but in practice I have found it just as good if not better to set alerts for volume cues and then look at the chart..........................
     
  4. ivob

    ivob

    Hello,

    Why do you think it doesn't work?

    I cannot judge at this moment whether it is fast enough or not but :

    1. I imagine this also depends on the timeframe I want to trade. If every millisecond counts it may be too slow. Actually from the little testing that I have done it's not too slow and no delay can be noticed.

    2. It depends on the calculations that are made and what you want the system to do. If for instance the system is programmed to give me a signal when a cup is formed then I can wait for the handle myself and trade manually. Of course if this works it can be improved later.

    3. It depends how it is programmed, if the system runs as a daemon, etc.

    regards,
    Ivo
     
  5. agpilot

    agpilot

    Ivo
    If my memory is right, I think John Murphy had played with this along with Metastock software several years back. (Murphy wrote several TA books and Metastock is a software charting program) I hope this is what your searching for... Good luck... agpilot
     
  6. rosy

    rosy

    perl/python is used for this exact problem at a several large trading firms that I am aware of. Basically, logic in perl and infrastructure in C++. You can use swig for optimization.

    That being said, a one man shop trading out of a retail account is no match against a well funded firm.
     
  7. I don't think it would be difficult to program in Tradestation Easy Language. However, with all the nuances of the market, visual methods can be developed also.
     
  8. nitro

    nitro

    This is a project I am working on the side because I find it interesting.

    If you know how to program and know a little mathematics, it would not be terribly hard to do this yourself.

    Make a correlation template for each of the patterns you want to detect. Convolve the signal with the template. If the correlation is high, the pattern is there. This scheme is robust even if the match is not exact since small changes in the signal are mapped into small changes in correlation.

    The only problem is that what you really want is a multi-timeframe version that can do it on all time frames at once. For this you need a set of wavelet "templates" or basis, and that requires quite a bit more sophistication on your part to implement.

    nitro
     
  9. Google for this - might be helpful

    Foundations of Technical Analysis:
    Computational Algorithms, Statistical
    Inference, and Empirical Implementation
    ANDREW W. LO, HARRY MAMAYSKY, AND JIANG WANG*
    ABSTRACT
    Technical analysis, also known as “charting,” has been a part of financial practice
    for many decades, but this discipline has not received the same level of academic
    scrutiny and acceptance as more traditional approaches such as fundamental analysis.
    One of the main obstacles is the highly subjective nature of technical analysis—
    the presence of geometric shapes in historical price charts is often in the eyes
    of the beholder. In this paper, we propose a systematic and automatic approach to
    technical pattern recognition using nonparametric kernel regression, and we apply
    this method to a large number of U.S. stocks from 1962 to 1996 to evaluate the
    effectiveness of technical analysis. By comparing the unconditional empirical distribution
    of daily stock returns to the conditional distribution—conditioned on specific
    technical indicators such as head-and-shoulders or double-bottoms—we find
    that over the 31-year sample period, several technical indicators do provide incremental
    information and may have some practical value.
     
  10. ready

    ready

    Prophet.com long and short term patterns found and searchs for paticular patterns
     
    #10     May 20, 2006