Fractal Analysis?

Discussion in 'Technical Analysis' started by youngtrader, Sep 1, 2007.

  1. Kuran47

    Kuran47

    "Can anybody explain what this is and how traders use it in the markets? Is it the same thing as chaos theory and what is that? Sorry about all the newbie questions but I would like to know more about these stratagies."

    :

    Market price distribution conforms to a Stable Pareto Levy distribution, not Normal / Gaussian distribution. One of the properties of a SPL distribution is that the short term is "predictable" but the long term is not (sensitivity to initial conditions). One great example of a non linear dynamical system (which is what the markets are) is weather systems and their forecasts. Ever notice how the weather forecasts are never given for more than 5 days, forecasts made 5-6 days ago are subject to changes, and never given for more than 3 weeks or 3 months or even 3 years out? And the forecasts are rough estimates? Weather systems are driven by aperiodic cycles of high and low pressure systems...

    Same thing with markets.

    Check out my blog for more info:

    http://gnomesden.wordpress.com/
     
    #11     Sep 4, 2007
  2. kut2k2

    kut2k2

    I agree with most of this but I don't hold out much hope for chaos theory applied to the markets. CT was invented by pure mathematicians, not by engineers or applied scientists who have to be practical in their speculations.

    Chaos theorists came up with the Alice-In-Wonderlandish idea of the Butterfly Effect: a butterfly flapping its wings in Brazil will start a series of events that will result in a storm in Beijing. Obviously those people have never heard of damping aka air friction. Show me the butterfly that can flap its wings and affect the weather, and I'll show you Mothra. :p
     
    #12     Sep 4, 2007

  3. Interesting ideas. I don't think fourier analysis has much to do with fractals.
    However, your site has an interesting approach regarding fft based prognostication, nevertheless.

    Do you ever overlay the actual data over your fourier predictions? Can you post some here? Would like to see how close they align.
     
    #13     Sep 4, 2007
  4. I don't have your raw data, but here's a quick overlay example of wide divergence in your july 24th predictions. Do you ever see overlays that align close to the predictions? It would be nice for your blog to follow up with correlations or overlays, of predicted vs. actual results. Also, I assume you know, ffts are very sensitive to taking the proper window of data, to avoid sampling error.
     
    #14     Sep 4, 2007
  5. Kuran47

    Kuran47

    "Interesting ideas. I don't think fourier analysis has much to do with fractals."

    :

    Yup. FFT has nothing to do with Fractals or chaos theory (I prefer to call it nonlinear dynamical systems rather than chaos theory.). But what it does is gives me a good look at short term aperiodic cycles.

    It is my opinion that the market is "predictable" in the short term, 3-5 days out or 1-2 cycles out. But it changes as the cycles are aperiodic. That's why I update it constantly as best I can, subject to my ability to access the internet as I travel.

    Regarding the July 24th projections, it was made on data ending on July 20th. I followed that post up with on August 5th, with the raw data outline to match the projections.

    The post on July 24th was made using the S&P 500 futures data, not the eMini S&P (though for some unknown reason, I posted it instead of the eMini's.). As I have not subscribed to the CME for the "big" S&P 500 (as I don't trade it) and so that the data you see is delayed. In general, both the S&P 500 and the eMini line up, though there are times when they do diverge. Most of the time, they do line up but might have different peaks offset from each other by a few hours or so . I used both data to get a sense of whether or not they are in line with each other and that allows me to "forecast" the future. :) Normally, I post the prognostication with the eMini's, though for some unknown reason, I posted it with the S&P 500 that day.

    Regarding this statement: "Also, I assume you know, ffts are very sensitive to taking the proper window of data, to avoid sampling error."

    Yup. That's why I had to modify it to account for this sensitivity... as well as for other stuff.

    I think with CT, as most people refer to it, gives you an idea of what the market is about, more of a market philosophy rather than a predefine set of rules, set in stone (another market philosophy is EMH, and its offshoot CAPM, etc.).

    My take on it is that the markets are "predictable" in the short run, as any changes in market conditions can have a substantial effect in the long run (sensitivity to initial conditions). Long term forecast are a waste of time. I think my analogy to weather systems, consisting of high and low pressure systems constantly rotating against one another, is reasonably accurate. Though there are seasonal factors in the markets, as it is in weather system (cold winters in the Northern Hemisphere, warm summers, etc.), it is the state of the market that is "predictable" in the short run and that is what I'm concern with.

    Just my $0.02....
     
    #15     Sep 4, 2007
  6. Profitunity is a service that uses fractals (at least, they were some years ago). I found little value in them when I looked at them. Don't really know if anyone is using this to make accurate calls, as I see little other than hype about this stuff
     
    #16     Sep 4, 2007
  7. Kuran47

    Kuran47

    here's 1 GIF files taken off the blogsite with the raw data overlaid on top of the projections.

    It seems as if I can only attach 1 GIF file at a time.

    In it, the red vertical lines denotes a high while the green vertical lines denotes a low. The red "horizontal" lines denotes a losing trade while the green "horizontal" lines denotes a profitable trade.
     
    #17     Sep 4, 2007
  8. A few years ago Bill Williams wrote a book on this topic, Trading Chaos and attempted to systematize trading using "fractals" based on his interpretation of certain aspects of chaos theory. He followed up with software and a second book, New Trading Dimensions. I've gone through it all and as I recall uses a collection of moving averages in different timeframes and several other indicators. FWIW, too many moving pieces for me and I did not perceive an edge.
     
    #18     Sep 4, 2007
  9. kut2k2

    kut2k2

    Bill Williams is one of those charlatans that jimrockford posted about. There's nothing about actually using chaos or fractals inside the covers of "Trading Chaos", it's all moving averages and fibonacci crap.
     
    #19     Sep 4, 2007
  10. Oh yeah, I did see some of those overlays. Only, I"m not certain which part was predicted and which is post data overlay. Is it only the period where the light blue continuous line is the prediction? Because that's only about a quarter of a cycle. How about like a whole month or longer overly, or are you saying only the short quarter cycle length is expected to be predictable before divergence?

    I pretty much see a lot of truth in your whether/prediction analogy.

    By the way, since a few posters are talking about the original post, fractal analysis. I've mentioned this several times, about the most objective and uncannily accurate approach to elliot waves I've seen (and I've seen a lot) was done using fractal generation.
    There is an article about it in ta of stocks and commodities, titled "fractal analysis."

    http://www.traders.com/Documentatio.../042007/Abstracts_new/Williamson/william.html

    His analysis called a top a while back ( in that article) and is looking for 25% down before this correction is done.
    I will be pretty amazed if his target is hit. And the greatest thing about it, is there is little subjectivity (although, I still had a few questions after conversing with him regarding methodology). Fractal enthusiasts might want to check it out.

    The other fractal approach is used to determine trending or trading periods
    using hurst exponents (but has little use in predictability of future IMO).
     
    #20     Sep 4, 2007