Random Chart Contest: RESULTS!

Discussion in 'Technical Analysis' started by jmcgraw, Dec 10, 2002.

  1. I didn't participe: I have absotly no intuition at all :).

     
    #11     Dec 10, 2002
  2. But were the ascii datas ? I'm interested to get them to make tests when I will have time to do so.

     
    #12     Dec 10, 2002
  3. JayK

    JayK

    Very interesting :)
     
    #13     Dec 10, 2002
  4. jmcgraw

    jmcgraw

    Go here:

    http://users.rcn.com/jmcgraw/

    and click on "raw data". Its all there.


    .
     
    #14     Dec 10, 2002
  5. jmcgraw,

    That was a really fun challenge!:)

    And using the amazing power of hindsight I can see how I missed the two answers that I did. Now if I only had a way to get hindsight before I trade.....:D

    Couple of suggestions for making the next contest even more realistic (even harder):

    1. Create positively autocorrelated volatility by letting the magnitude of the random increments be an auto-correlated random variable itself. This would be enough to fool the simple approach that I used in the first contest. Also chart 6 had strongly anti-correlated volatility, so that may be why it was so easy to pick as a fake.

    2. Double-check the OHLC data: a number of numerical data sets had anomalous records where the close of the day was higher than the high of the day (or lower than the low of the day).


    Thank you for such an excellent challenge and for the interesting analysis of everyone's votes. I look forward to the next contest (where I will perform even more poorly if you implement the preceding suggestions.)

    Trade well,
    Traden4Alpha
     
    #15     Dec 10, 2002
  6. Quiet1

    Quiet1

    At least then I could add "lab rat" to my CV.

    Just don't make me run round one of those treadmill things or go looking for cheese...anything but that...

    :)
    Q1
     
    #16     Dec 10, 2002
  7. Chart #11 will be random. How do I know? Just look at the pattern. Random charts always occur in groups of 2.

    Chart 1 and 2...random

    than all real charts

    Chart 6 and 7 random

    so after 1 random chart #10 a second random chart #11 must follow.
     
    #17     Dec 10, 2002
  8. But not for long.. :D
     
    #18     Dec 11, 2002
  9. BKuerbs

    BKuerbs

    Some nit-picking: an auto-correlated "random" variable is no true random variable, just a function.

    Regards

    Bernd Kuerbs
     
    #19     Dec 11, 2002
  10. bobcathy1

    bobcathy1 Guest

    I am amazed I got 2. I just guessed at random.
    I could not see anything amiss in those charts, they all looked real to me:eek:
     
    #20     Dec 11, 2002