Mathematically Predicting the Future?

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

  1. fatrat


    Attribute those quotes correctly. I didn't make that nutty statement. :)
    #31     Oct 28, 2007
  2. I want to see if Marks systems can serve as a useful filter for my current sub-systems.

    So, I'd like to run Marks "signal files" in combination with my sub-systems to see if performance is improved. Thats all, no big deal.
    #32     Oct 28, 2007
  3. Dood you gotta chill out. Done! Stratcat..err... Fatrat rightly removed from the quote in previous post.
    #33     Oct 28, 2007
  4. I think most people are in denial about their ability to predict the market. I am a believer in a semi-strong efficient market hypothesis. This is only magnified by trading costs (slippage, brokerage/exchange fees, trader errors, and risk factors such as RiskofRuin)

    No matter how much people here claim to regularly outperform, the fact remains that it is believed that about 95% lose their money during leveraged trading. I am willing to bet a goodsized chunk of the remainder are barely profitable.

    On ET, many who make claims, do so without proof. I have seen considerable evidence that people like ProfLogic, who claim to be profitable, do so by ripping off others. I posted a PM about someone who said he was nothing more than a flim-flam artist. WHen I challeneged him mano a mano, he would only do so if I wouyld bet about $50,000.

    People like Jack Hershey love and crave the attention. But when challenged to put up, they have many excuses. I challeneged him mano a mano, but he declined, saying his statistics would be "too good for my standards." Of course, the only real test of his prowess seemed to be a 26% loss and causing his early departure from testing.

    I put up a clear challenge to TA lovers. After over a 150 pages of posts, the claims from TA believers basically boiled down to:

    --- "of course classic TA doesn't work. You have to know how to modify them to your purpose
    --- "the numerous studies showing that TA has no predictive value, using many diff indicators over many years" are pointless. Just question one or two of their points (even though they survived industry/academic review), nad claim the whole thing is pointless. OF course, TA beleivers offer little in return
    --- the performance of CTAs who generally favor TA when monitored, year in and year out, remains pathetic. And simple blind, index funds continue to outperform most stock advisors/funds most years.
    --- the performance of "Advisors" on places like Timertrac, Hulbert, and other independent monitors remains very pathetic. Many/most of these claim TA bases.
    --- In general, the strongest responses from these people are insults, ridicule, retors, anger, "works for me" but VERY little clearly defensible evidence. I doubt the claims of most people who shy away from

    Recently, the guy who started the forum on the back of ET earlier this year, was doing relatively well trading multiple "best" FXauto systems (forex). The appeal was high and many others started. But over the last few months, he is now down about 50% overall and trading none of them (as of this week)

    So in answer to your question trading-wise, I doubt it...

    But a few make money... A few
    #34     Oct 28, 2007
  5. Kuran47


    Not my arguement. Fama's:

    Fama, Eugene (1970). "Efficient Capital Markets: A Review of Theory and Empirical Work". Journal of Finance 25: 383-417.

    That's what I'm disputing. You can fit any kind of distribution on market movement, including Normal / Gaussian, but what I'm saying is that the best fit is SPL.

    As for the load of crap you've deleted, I've attached a GIF file showing the Hurst coefficient on S&P 500 futures. If the markets price movements are random, it'll be exactly 1/2, 0.50. it's not.
    #35     Oct 28, 2007
  6. Kuran47


    My basic arguement is that market prices do not change as if it were a random walk (this can be proven mathematically - via Hurst Coefficient, Kurtosis, Skewness, etc.).

    Just because something is unpredictable does not mean it is random. Since I'm not very articulate on this point, I'm going to let someone else make this point for me:

    "Randomness versus unpredictability

    Randomness is an objective property. Nevertheless, what appears random to one observer may not appear random to another observer. Consider two observers of a sequence of bits, only one of whom has the cryptographic key needed to turn the sequence of bits into a readable message. The message is not random, but is unpredictable for one of the observers. One of the intriguing aspects of random processes is that it is hard to know whether the process is truly random. The observer can always suspect that there is some "key" that unlocks the message. This is one of the foundations of superstition and is also what is a driving motive, curiosity, for discovery in science and mathematics.

    Under the cosmological hypothesis of determinism there is no randomness in the universe, only unpredictability.

    Some mathematically defined sequences exhibit some of the same characteristics as random sequences, but because they are generated by a describable mechanism they are called pseudorandom. To an observer who does not know the mechanism, a pseudorandom sequence is unpredictable.

    Chaotic systems are unpredictable in practice due to their extreme dependence on initial conditions. Whether or not they are unpredictable in terms of computability theory is a subject of current research. At least in some disciplines of computability theory the notion of randomness turns out to be identified with computational unpredictability.

    Randomness of a phenomenon is not itself 'random'. It can often be precisely characterized, usually in terms of probability or expected value. For instance quantum mechanics allows a very precise calculation of the half-lives of atoms even though the process of atomic decay is a random one. More simply, though we cannot predict the outcome of a single toss of a fair coin, we can characterize its general behavior by saying that if a large number of tosses are made, roughly half of them will show up "Heads". Ohm's law and the kinetic theory of gases are precise characterizations of macroscopic phenomena which are random on the microscopic level."
    #36     Oct 28, 2007
    Simples likes this.
  7. During the winter months will you go out in your summer cloths?

    Some may argue the market has no predictable series or bias. If that market doesn't why trade it, analyzing 1000's of stocks and commodity charts and various markets, there are markets stuck in winter and summer months.

    Dress accordingly.
    #37     Oct 28, 2007
  8. MarkBrown


    nice chart and post! mb
    #38     Oct 28, 2007
  9. MarkBrown


    oh! you know now that i understand how you will use it i have something much better than a system i think. let me get it in a file and i will send it out in that format you want.

    it will be a trade direction filter. mb
    #39     Oct 28, 2007
  10. Re-read your response to fatrat earlier in this thread. You state that for the markets to be random they must be Gaussian. This is what you wrote, and what I am basically saying is you should go read a stats book, because quite frankly that statement is wrong.

    I don't give a toss how you think the markets behave, or get entangled in the semantics of randomness and unpredictability or the emergent behaviour of complex systems, and I am not passing judgement on your other views (you can think whatever you like bro - its a free country), but your "Gaussian" statement is plainly wrong regardless of its origin.

    You see, I couldn't give a shit if you trade standing on your head whilst ticklin' the iris, or you apply Chern-Simons theory. At the end of the day only one thing matters. Can you make money?
    #40     Oct 29, 2007