Will this work?

Discussion in 'Automated Trading' started by StarfishPrime, Apr 28, 2010.

  1. Let me start this off by saying something about myself. I'm a pretty smart guy and have been searching for a profitable method for the last three years, but I've been unsuccessful so far. Not because I've actually lost that much money, but because I haven't quite applied myself to theories that could make money. Right now I'm a full-time student in college and I've been interested in the markets since I was 16.

    I remember how it all (auto-trading ideas) got started when I had just turned 17: I downloaded some version of MetaTrader from some random forex broker and just kept looking at what all the indicators could do. I got to the Ichimoku Kinko Hyo indicator and thought, "wow, this almost cuts through the crap the other indicators generate." Then I quickly browsed through the docs and within a few hours, I had my first auto-trading strategy with super simple rules. Buy when the one line was above the other and sell when that line is below the other (I don't remember the exact names of the signal lines). Quickly after that I had to try the moving average crossover, just to see how bad it performed. It was terrible, as expected. Then I had an idea, why don't I just flip the signals? I autotraded the reverse MA crossover for two weeks on YM. Some days were good and some were bad. Overall, not good enough to go live.

    Since then, I've experimented with almost the entire spectrum of trading ideas available to the retail trader from breakouts, support & resistance, and chart patterns all the way to looking for arb opportunities through forex "rings", spread trading (energies, bonds, including foreign markets), and reversion to the mean type trading.

    Maybe it was the YM "fading" autotrader I created back in the day that spurred me, but I always come back to RTM trading because it seems so predictable.

    Due to not ever having enough time (cause I'm trying to have a life while studying full-time in addition to trading activities), I haven't ever fully created a viable trading system, but I just need to bite the bullet and develop something. I know I'll be able to code it if I put my mind to it, it is just a matter of actually getting far enough into the development process to realize that I have a potential winner on my hands.

    With that introduction out of the way, I present to the community of Elite Trader the basis of what I believe will work as a profitable trading system. My only two questions are: do you think I have enough to work with? And also: should I go for it?

    The following graphs show four things: the changes in the price (deltas) of a synthetic instrument I've created for both an extended period and a close up of a certain time period, as well as the distance said instrument is from an "average of averages" of sorts for the same two periods.

    If anyone has any additional comments, I'd love to hear them.

    I will have to post these pictures one at a time.
  2. Somehow my message got held up...
  3. More data:
  4. Last graph for now:
  5. corbel


    Do it

    That may be anticlimactic after your long post, but it comes down to: do it

    But, make sure you don't do it wrong. At first, you want to be going after pennies. If you can consistently profit a penny each time, then you're successful. Then do it several million times a year. Don't set your goals super high, like it appears that you may be going for. In the market, the intelligent, emotionless, patient people always win.
  6. LeeD


    "Big-picture" charts looks very much like what you'd get from a GARCH-type model. Close-ups show that you may have to take care of drifting mean.
  7. IMHO, It won't work.

    The mean is defined on collected data.

    There is no such a thing as a "reversion to" the unknown (future) mean.

    The "reversion" is an illusion created "a posteriori", by the fact that you
    compute averages on known data.

    (It's conceptually equivalent to making previsions or directional assumptions)