Mathematically Predicting the Future?

Discussion in 'Automated Trading' started by MarkBrown, Oct 27, 2007.

  1. MarkBrown


    Mathematically Predicting the Future?

    Has it Can it be done? As I watch, look and read everything available about trading I have to wonder where did all this junk come from. More than ever I think many people talking about trading using this and that method. Meaning the methods are not purely mechanical, I think we have and are going backwards.

    Man in not progressing forward, it seems we are just becoming a Byzantine trading society. We just jewel up a moving average and get some guy on a video saying had you done this and had you done that you would have - could have made this on that trade. Where's the beef? Hey just forget all that stuff tell me where to buy it and where to sell it etc.

    Mathematically is it possible, can man define the markets and surround it with a mathematical fence? Pull the market around on a leash like you were walking a dog? Sophisticated software with Ph.D.'s and global knowledge and light speed communication all at our fingertips. Why is it that the perfect trading method does not exist or does it?

    Inquiring Minds wanna know, your thoughts. mb
    .sigma likes this.
  2. Any update on OddBall system performance this year?
  3. MarkBrown


    i plan to fully update oddball soon and fully/freely disclose many more versions i never talked about before. i think it will be the best stuff ever. mb

    ps i have a day trade version too.
  4. Do you have "signal files" reflecting EOD trades for your systems:

    Russell 2000 Index RL
    S. & P. 400 MD
    S. & P. 500, Day Session SP

    In order to run trade history of systems on my software, need format like:

  5. MarkBrown


    i will be able to accommodate soon i am trying to stay focused on some stuff right now. maybe in a week. i really have a good system on the sp400 funny you asked that. mb
  6. Sounds good, stay in touch...
  7. MarkBrown


    do you just want the trades or how about i just give the system details. mb

    like gimmie a example of the expected trade output format.
  8. I just want to run the EOD Buy/Sell signals of your system on my software to see how the various systems performed historically.

    (System details are your property, of course, which I would purchase after running systems and combining with my own sub-systems etc.)

    Below is example of EOD System Performance Summary from my software:

    Trades: 424 (71.93% Winners) 17.52/year
    Ave. Gain Per Trade: +0.74%
    Largest Gain: +5.14%
    Largest Loss: -17.34%
    Max. Drawdown (peaks): -28.51% on 10/19/87
    Max. Drawdown (entry): -27.33% on 10/19/87
    Total Gain: +2087.89%
    CAR: +13.60%
    Percent Invested: 21.47%
    Percent long: 21.47%
    Percent short: 0.00%
    CAR While Invested: +48.62%
    Ulcer Index: 3.62
    Winning Years: 92.00%
  9. Kuran47


    MB wrote:

    "Can it be done? "

    I think it is possible. Depending on the degree of accuracy , I think it can be done.

    After all, we are predicting the weather, aren't we? Though we use satellites in weather prediction to get an instantaneous picture(we use charts about the same way meteorologists use satellites), at the end of the day, it's still dependent on number crunching in the weather forecast models. Which is not a bad analogy as both the weather and the markets are nonlinear dynamical systems.

    Notice how weather forecast are accurate to just 3-4 days out? And you don't see forecast 3 months or even 3 weeks down the road? At you don't see forecasters saying that at 2.54 pm EST at 39th and Park, the temperature will be 84.7 degrees with winds at 11.2 miles an hour from the west, with barometric pressure at 29.543 inches? Most forecast are rough: a high pressure system is moving in, pushing out the low pressure system responsible for the rain of the last few days, expect clear sky with some clouds, winds at about 10 mph out of the west, etc.

    Same thing with the markets. It has to be rough to be accurate, can't pinpoint the exact prices or time, to generally give you an idea of the market structures will look like over the next few days.

    I've attached a GIF file showing something that is analogous to weather forecast for the markets. It's in TS, with the forecast on the bottom with the subsequent market action on top. Disregard the time the vertical bars denotes. It's designed to be "accurate" over the next several days, like weather forecasting, not to be accurate to the hour. The red vertical bars picks a day for highs, yellow for lows.

    MB then wrote:

    "Mathematically is it possible, can man define the markets and surround it with a mathematical fence? Pull the market around on a leash like you were walking a dog?"

    I don't think so. The markets as a whole is greater than any single man. As much as we benefit from weather forecasting, no one is yet capable of controlling the weather (we can seed a cloud to induce rain, but we can't create the clouds to carry rain or tell it to carry it to where the rain is needed.). Same thing with the markets. We can take shelter from the storms if we know it's coming but we can't control it.
  10. Kohanz


    If by predicting you mean to be able to perfectly predict where prices will go, I don't think we will ever reach that stage. It is too complex of a system, with too many random, changing variables that cannot be characterized using any tractable math we have available to us today.

    The previous poster spoke of predicting the weather, but actually our weather prediction capabilities are not that accurate when you take a look at things - they can't even tell us with certainty if it will rain a day from now or not in a lot of cases.

    Because we can predict things with *some* accuracy, there exists the opportunity for profitable systems, but perfect or near-perfect prediction will always remain out of reach, IMHO.
    #10     Oct 27, 2007