Mechanical Trading ... Does it work?

Discussion in 'Trading' started by Brutus, Mar 6, 2002.

  1. Brutus


    I am pondering whether to try to develop mechanical trading systems or focus my trading on discretionary trading. Mechanical trading is easy to execute and there is not much thinking involved. That is all done beforehand in developing the system. From this perpective mechanical trading seems like a good idea. If you end up losing money, you can ultimately blame the system as being bad and create a new one. Discretionary trading on the other hand requires mastering of your emotions and thinking when your executing trades. Discretionary is not brainless following of rules.

    My problem with mechanical system trading is my belief that all mechanical systems will fail if given enough time. The system might work for a year or ten but then when it fails you lose big.
    I have heard of this type of thing happening. This then leads me to the conclusion I should just focus on discretionary trading rather than developing mechanical systems.

    I would like to hear the opinions of the successful mechanical system traders as well as those of you that rely on intuiton.

  2. stevet


    can i have some of what you are on - it must be great!
  3. tntneo

    tntneo Moderator

    this topic has been beaten to death. but it is always fun to see a thread get crazy on this very issue.

    0f course mechanical trading works. You probably can't afford day trading mechanical system development. Well, at least not at the level of what works.

    Or course most traders are not using mechanical trading. And they don't need to.

    Of course mechanical trading does not work forever, you need to change system from time to time. Nothing works forever. (there are strategies which work, then don't, then work again. they have cycles too).

    For individual traders (ie not hedge funds), learning about mechanical trading is very interesting. because it teaches to have a method, follow it with discipline and recognize when trading stops working.

    There are many ways to trade and make money without mechanical systems. so you really don't need it.
    After using mechanical systems last year, it improved my trading (oddly enough, since I was mostly watching the machine then). Nowadays, I would be in the discretionnary camp. but maybe it's misleading because my trading is very systematic. however, it's not a machine doing it and I even hardly use all the software tools I used to.

    I would really recommend trying a mechanical approach (not necessarly actually automated) but just let the rules make the decision for you. After a while, you don't see the market the same way. Also, you get a feel for when the trade won't work (which is nice of course).

  4. TNT got it right, we have people doing it, and making money...high cost of development (plus the cost of keeping several traders to monitor exits, etc.).

    Will it always work, doubtful. 1 in 50 may do well IMO. It used to be 1 in 100, ....
  5. alain


    to TNTneo

    You mentioned a very interesting point. If I understand correctly through trading with a mechanical system you were able to get a feel for the markets. And then after a while you were able to trade by what you feel. more or less...

    If that is ok with you I would be interested in a more detailed description of how you went through this process and what the key elements where in your process of trading.

    thank you very much in advance

  6. Dearest Brother Brutus,

    I prefer discretionary trading, which is rules-based... at first glance, what I have just said sounds like a paradox, but it isn't... it is discretionary in that I choose with my gut which trades to take and how to manage them... but at the same time, I utilise a list of rules to guide me in my trade selection and management... the more the rules tally up, the better... but the rules don't trigger my trade or get me out of a trade, they simply provide some degree of structure to my highly subjective discretionary assessment of a whole bunch of variables.... such a discretionary assessment is based on experience and cannot be put down in any meaningful way on paper, since the discretionary assessment is a manifestation of my gut feel... and, of course, gut feel by its very nature has both rational and emotional constituents, making an analysis of it unique to the personality traits of a given trader...

    With the love, warmth and fondness for you that symbolises the spirit of comradeship amongst the Trading Brethren,
  7. jaypaul


    Not all trading systems simply fail. It's more a matter of how well they adapt to changing markets.

    Some tips:

    1. The goal of designing a system is not only to "trade profitably", but more to "adapt to changing markets". If you want a system that can adapt in the future, first make sure it can adapt to the past, that it is statistically stable over many years of history, in different markets, different situations and in trading different equities/commodities if possible.

    When backtesting, force your system to make a large number of trades, but don't let it concentrate trades or equity in one type of market, or at only certain times or situations. Force it to adapt to all markets and situations. Price behavior for a single stock or a single situation can change much faster than most systems can adapt. There's a lot of market history available. Why not consider it all?

    2. Everyone is analyzing and trading with price and price-based indicators. Instead, consider analyzing more obscure, under-appreciated data like time & sales.

    3. Don't start out with complicated methods like neural nets or by tweaking buy/sell thresholds. Design simple statistical tests to explore your data, form a quantitative hypothesis to explain some aspect of future price changes, test the hypothesis, try to understand the result, then discard or revise the hypothesis.

    4. Write your system in MATLAB, or with some toolkit that gives you a rich set of numerical processing, statistical and visualization tools.

    5. Be prepared to spend months or years doing this, unless you are mathematically gifted, or if you can blend an objective system with subjective, human interpretation.

    Good luck!
  8. tntneo

    tntneo Moderator

    OK, I don't want to make a big thing out of it. the learning curve and experience is probably very different for each trader.
    for me, I could much better understand market dynamics (and what my mentors were telling me) using mechanical systems.
    this is probably because a mechanical approach detaches you from the actual trading. I became more neutral (no greed , no fear).
    In that states it was easier to see the waves of buyers and sellers and how they move the markets.

    When you trade yourself, you 'want' the market to do this or that. it's a dangerous proposition and a lot of the trader psychology stuff is about removing yourself from the trade.

    I guess, you wanted to know more about the market dynamics itself. again a mechanical approach forces you to make formal descriptions of your belief system (you trade your belief not the market). So if your belief is wrong you see it pretty quickly. but then you also see what the market is really doing 'against' your belief.

    I will just take one example to illustrate my point. (don't expect me to explain you how I see the market dynamics itself, you have to find out by yourself. although many have written about it in books).
    GAP is the example. A gap is a break out of a previous short term trend.
    normally, a break out system would buy a gap up. (that's the belief : break out up, is long, buy it).
    however, if you mechanically do that, you will quickly realize that the market usually trade AGAINST a gap up. so while you buy, most short term traders will short and fade you.
    there you have an example.
    the better approach for many months now is to fade break outs. I would not say it is always the case, but odds favor this.

    when it does not work, it's also a very important piece of information.
    mechanical approach shows you it is as important to know you are right, as wrong.
    a bad trade is sometimes a great opportunity in the other direction. not always, but depending on your rules it might be consistently very interesting to reverse your bad trades.

    you can't really improve trading unless the trading is rules based. otherwise how to get back and study what worked and what failed.
    that's why imho mechanical trading (even without automation or software) is very important while you learn. you must do always the same thing to know what works and what does not.

    I remember another thread where we mentioned how many traders lose (!) because they constantly change strategy and chase the latest 'trick' or 'tip'.

    now, my trading is discretionnary in the sense that I do follow rules but in a more flexible way. I don't take all the trades etc.. all the things you must NOT do when you learn or test a system. but it is sure more fun now.
    you can't skip the learning though.

  9. BruceF


    I believe we trade our own beliefs about the markets. If you assume that is true, then you answered your own question.
  10. Magna

    Magna Administrator

    Damn, what a loose usage of the English language (jes' kiddin'). :)
    #10     Mar 6, 2002