Challenge: Real or Random?

Discussion in 'Technical Analysis' started by jmcgraw, Dec 3, 2002.

  1. jmcgraw



    I have written a little program in Excel that creates charts in accordance with specified volatility settings (I adjust these to simulate stocks with different trading characteristics) but with a 50/50 probability of either opening up or down on any given day. (The close price is a random draw between the high and the low of the days prices) So the trends and patterns are random. (Well at least pseudo random, since computers cannot create truly random numbers. But as far as I can tell the data does not trend in any specified direction nor are there any common patterns... In fact at a glance they look pretty much like your average stock chart)

    So, to summarize. The volatility is random within specified boundaries, and the price change is random within specified boundaries. No one claims that trading activity is 100% random, everyone concedes that boundaries exist most of the time. However the question is whether the price movements are themselves random or predictable within those boundaries. (I hold the belief that they are somewhat predictable)

    The Contest

    As I said, I do believe there is predictability. So I would like to pose a challenge to all the traders on this board. I have set up a website (Don’t worry, there is absolutely no commercial content, I am NOT trying to sell anything!) where you will find 10 stock charts some of which are random, and some of which are real historic charts. Each chart covers about one year's trading, and they are all daily bar charts. I am curious (and I am sure it will be interesting for everyone here as well) to see how well anyone who is willing to participate can separate the phonies from the real thing. So here are the guidelines:


    1. Visit the following website, where you will find the 10 charts:

    2. Reply to this message with your assessment of which charts are random and which are real.

    3. Specify what methods you used to make your decisions. (Intuition, traditional TA, Elliott wave, etc)

    4. Specify how many years of trading experience and/or studying the markets you have.

    5. Since some of the charts are real, it may be that if you are an older trader you could recognize them. (I have only included charts that are over 8 years old) If this is the case, please do not reveal your certainty until the answers have been revealed.


    I will reveal the answers in about a week, or whenever I stop getting replies. I will then post the scores in raw form, as well as broken down according to trading experience and analysis methods. I hope this will prove to be interesting.
  2. dottom


    Interesting way to have some fun.

    Can you also post the OHLC data in CSV format? Some types of analysis require the raw data.
  3. You should also place the charts in candlestick view .. might help some of us.

  4. jmcgraw


    I will post tab deliminated text files of the data as soon as I can. I just need to filter the data so it all looks the same and there are no giveaway clues of which charts are which.

    I will also try to get some candlestick charts up as soon as I can.
  5. great great post ! i love it !! posts like this make et worthwhile--------

    ok here is my ascessment:

    1. random

    2. real

    3. real

    4. random

    5. real

    6. random

    7. real

    8. random

    9. real

    10. random

    now, of course this is pending no trickery involved---like they are ALL random or ALL real.


  6. forgot to add background info per request. check out my master interview currently posted at under the "dave's wave" tab for bio/interview. cant wait to see the answer !

    surfer:cool: :)
  7. dottom


    In the random charts, how are you determining the magnitude of the change? In other words, you are using 50/50 random for the sign of the next day's close, but is the magnitude of the move based on brownian motion or your own custom derivation? If it's just as likely to change 3% as it is 0.5% then it's not true random walk.

    If you didn't use brownian motion to represent a true random walk, but used the same principle (that it's less likely that a larger magnitude move would occur than a smaller one) then it could be "close enough".
  8. acrary


    ok, here's my guess:

    Random: 1,4,5,6,10
    Real: 2,3,7,8,9

    Method used: Just looked at open-close gaps.

    Background: Retired derivatives trader after 14+ years at a investment bank.
  9. jmcgraw



    I dont know what brownian motion is. But this is what I do. I set up levels of probability, then set a volatility for each probability. So an open gap of 0% to 0.5% might have a 80% probability. A 0.5 to 1.5% might have a 19.9% probability, and a 1.5% to 8% gap might have a 0.10% probability.

    I run an anlaysis on actual stocks to get ideas for realistic probability/volatility settings.

    Do you have any info on brownian motion that I could look at? Is my method "Close enough"?
  10. maxpi


    Real: 1,4,8,9, rest random

    Method: Quick look and guess

    Years trading: 3 research years, very little trading.

    #10     Dec 3, 2002