What is mechanical trading all about?

Discussion in 'Strategy Building' started by OPC, Mar 8, 2003.

  1. Interesting question. If the markets were random then no trading system could be built to perform but this is hotly debated.

    I run a small fund and if I were to give a simple example of a mechanical system it would be this: on a 15 bar chart, buy at the highest high of the last 25 bars and short at the lowest low of the past 25 bars. Year over year it will pull a profit with a small average trade and it will rely somewhat on outliers.

    There are many other ways to build systems anywhere from quantifying trends and getting on boards to simple breakout methodologies. Your platform is the best way to explore, there is ample material to work with out there. Hope it helps.
     
    #11     Mar 10, 2003
  2. The mojo or the instinctive objectivity is different with trading and system coding.

    The intuition for a system trader is finding a good concept and integral factors to make a more profitable and tradable system.

    The intuition for a discretionary (semis like signal pickers) are more inclined to real-time pattern recognition.

    Just because you're a good trader, it doesn't mean you're a good system developer. But being able to do both does help.

    The most important thing is to have a flexible market feel.
     
    #12     Mar 10, 2003
  3. OPC

    OPC

    ET and WDGann:

    I would not say that wind direction is completely random, in the same way I believe that markets are not completely random.

    What perhaps I am trying to convey is that we have an interesting phenomenon nowadays. With so many funds in the commodity and equity markets, there is a different element behind the market dynamics: price is not driven anymore solely by the human psychology, as we have the mechanical factor, i.e., the interaction of a multitude of trading systems generating price action.

    The end result would be something like this: the systems will be always dealing with cycles of contraction and expansion (agreement or disagreement about market direction). On the other hand, the event which will spark a contraction or expansion will be purely random in nature.

    So, I dare say that all mechanical systems are based on three principles:

    1 - Tracking expansion/contraction cycles;
    2 - Quantifying these cycles in different time dimensions for scaling in and out purposes;
    3 - Risk management.

    Now, as I have never traded mechanically, you are allowed to disagree. :)

    OPC
     
    #13     Mar 11, 2003
  4. Examples for futures funds:

    Long term trend following (daily data):

    Go long on next open if close > average (close,80) + standard deviation (close,80)

    Exit long on next open if close < average (close,80)

    Opposite for shorts.

    Or the turtle type rules:
    Buy stop at 20 day high, Sell stop at 20 day low, if in position go flat at 10 day high/low.

    Short term volatility breakouts (esp stock index futures):

    Take hi-lo range for last 3 days:
    Go long open + x% range
    If entry today, stop at today's open
    vice versa for shorts
    Exit on tomorrow's open if profitable, otherwise stop and reverse on a new signal

    x% commonly around 30.


    There is a real lot of money being run on these short term range breakout systems. With some careful research you could probably work out where most of the stops are being placed (ie what figure to use for x%), and then front run them to scalp a point here and there.

    e.g. if the S&P makes a drive from the open and moves enough to trigger some of these stops and quickly reverses, look to get set as the market moves back near the open for some follow through as they get out of their position.
     
    #14     Mar 11, 2003
  5. Good comments. I would argue that systems are a result of psychology and thus psychology still drives the market. However, systems will make the environment move faster be more competitive. Hedgefunds do move the market more than any other demographic though, I feel.

    It has been my experience systems built on cycles break down as there is no way to predict the length of a cycle. Hopefully risk management is incorporated. :)

    You are welcome to read a small book I wrote at TradingSystemDesign.com if you want more of a trader's perspective.
     
    #15     Mar 12, 2003
  6. I agree...

    Actually, I tend to think of the market as semi-random. It is because we all look at the same information whether it be a chart or a news. Most of the traders who hears a news knows whether the news is good or bad. When we see a breakout resistance, we know the anticipation of the direction of the market when it breaks out. In that sense, the market is not random.

    We hear and view the same information. But the way we interpret the specific pattern or news is different. The way we trade is also different too. There are scalpers, swingers, wave catchers and others and the stops and exits are going to change. In this sense, the market is close to random.

    One of the important factors with randomness of the perception is to look at the time frame. Scalpers use Tick to 1+ min. Wave and Momo(Momentum) traders use 2-5+ min. Swingers use 5+ min. Charts. I'm writing all this but everyone practically follows multiple timeframes to catch an edge. Still, this is important thing to consider.

    Also, another factor is that what is bad for one trader is good for another because of the Zero Sum nature of the trade. I remember George Soros saying that trading is the battle between each individual's philosophy. At sometimes, one side is winning, at other times they lose. It's something that we can't deny.

    So...

    When making a system, you need to have the objectivity of the nature of the trade and the conceptual edge defined. That becomes the building base of the system's "philosophy". From that base, we can expand and integrate the system itself. Also, noticing the objectivity causes the coder to identify the weakness by relating to the other side(s) to make a non-correlating system to cover the drawdown periods.

    Just my 2 cents.
     
    #16     Mar 12, 2003
  7. Nice...
     
    #17     Mar 12, 2003
  8. There is a misconception about randomness. If the market was random it would be too easy. I repeat what a statistician say: randomness means GLOBAL CERTITUDE, you cannot predict the NEXT occurence of an event but you can predict that it will occurs within X occurences with as high as probability as you want. And a random serie is in fact called a martingale mathematically and there is what is called a compensation law that permits you to win THEORICALLY but PRACTICALLY THE CASINO will emit RULES to prevent you to win for example by obliging you to bet a minimum and limit it to a maximum, and in stock market the equivalent of these limit rules are the fund account, the deposit, the margin calls ... and the non-randomness of market but rather a law that maximises volatility ! And some french econophysicians have just studied last year that statistically in an economical system, the higher the volatility the higher the gain for the big ones and the higher the probability of ruin for the small ones.


     
    #18     Mar 12, 2003
  9. OPC

    OPC

    ET:

    That's the random factor I was talking about. Hence risk management becomes a key component.

    Thanks for the book. I guess it's what I needed to grasp some mechanical concepts. I will read it as soon as I can.

    With regard to the principles I outlined before, that's the way hedge funds seem to influence price action, when considered collectively. It would appear that unless we have a range expansion they will not commit more funds into a given market or trend.

    OPC
     
    #19     Mar 12, 2003
  10. OPC

    OPC

    WDGann:

    Thanks for your comments and expertise. For the time being, however, let's say I will limit myself to grasping the concepts of mechanical trading so as to better understand market dynamics.

    OPC
     
    #20     Mar 12, 2003