The one thing I never understood...

Discussion in 'Automated Trading' started by Corey, Oct 27, 2008.

  1. Corey


    People always seem to be looking to develop the holy grail system. They start simple, and make the rules more convoluted as they cover corner cases ... and eventually the thing is a jumbled lump of spaghetti that doesn't even make fundamental sense.

    So here is a little bone I am throwing out there ... chew on it. I am not saying it is right or wrong ... it just seems to work for me.

    Instead of trying to write the holy grail indicator, why not spend your time identifying the current market condition? What does that mean? It means, write a system that identifies whether the market is favoring a trend following system, a mean reversion system, a fading system ... et cetera.

    Then, and this may be blasphemy, trade a very, very simple system whose bias is in line with the current market condition. Stop trying to write the perfect algorithm that can trade 100% of the time, and start writing the perfect algorithm that can trade 10% of the time. Then, when that time comes, trade it!

    Diversity, then, would be achieved by trading multiple (ideally non-correlated) assets.

    You see, this is why fund managers outperform over a short time period, then fade back into obscurity. It isn't because they had any insight ... it was more that the market aligned to their bias. Taleb's "Fooled by Randomness" and all that.

    So if you can identify which bias the market is favoring ... you can trade that strategy...

    Or maybe that is all just a bit too obvious...
  2. guauld


    you are so smart aren't you, well let me put you in your place

    what you are talking about is switching your strategy between mean reversion and trend,

    but you are so foolish you don't understand that while the market is switching you are entering draw down, and if the switching takes 10 days or so

    you are already -30 % in your portfolio

    it is OK, when you get burned you will learn

    as for you waiting and waiting for good setup and days passing you by, that won't make you rich

    I have a method that trades every single day no matter what market does, trends or chops

    makes money always

    put that into your smart pipe and smoke it buddy
  3. identifying the current market condition i call it CMC. is a great ability to develop and any system will have times when it produces unfavorable results. during these times we should not trade at all. a decent system should have setups at least on a daily basis.. i dont know what that guy is talking about
  4. Corey


    Given that you know nothing about my trading systems, my position sizing, or my risk management, you seem rather willing to speculate as to how long it would take me to switch systems or how much draw down I will incur. A rather audacious claim, don't you think?

    This does not require waiting and waiting for the perfect setup, it requires identifying the current market condition and the associated system that maximizes expected return in that condition.

    When there is no wind blowing, do you take out your sail boat? The same goes in trading: why use a trend following system if no trend exists (given a specific instrument and time frame)?

    I won't be so bold as to speculate exactly what your system is (or whether it is automated at all), but I will speculate that you have a mechanic that alters your parameters, effectively moving your from trend following to mean reversion and back. Or perhaps it stops you from trading altogether under certain conditions ... which is exactly what I am recommending: only trade when the conditions are good.

    I would like to point out, however, that your post doesn't have much credence when your account was just recently registered and this is your first post. If this is truly your first post, why so abrasive? If it isn't (which I would put money on), why hide behind an alias?

    Ultimately, what I am trying to point out is that sometimes it is better to find something that works in a corner case, and identify when that corner case will occur, than to try to find something that works in a broader sense. If you can come up with a tool box of these specific cases and strategies that can be extremely effective in them ... why not trade that?
  5. Im pretty sure he was just putting you on
  6. tommaso


    In my opinion, you are right Corey. There is no strategy which would work under any condition and any security.

    Strategies need to be calibrated both on market and financial instruments.

    And even "good" strategies should have the capability to take into account changing market conditions, or else they may stop being good very soon.

  7. Corey


    So I guess the question then becomes, "with what degree of certainty can be identify the current market condition, and are there any indicators that can help us identify when that condition is changing?"

    If there are no leading indicators, are there lagging indicators we can use that will help us identify the transition quickly without us incurring too much draw down while we use an inappropriate strategy? Perhaps this would tie in with our risk management -- given that each strategy should work in a given environment, if it stops working, it should trade smaller -- the market is identifying that the bias is not there.

    In which case, can we just have all the systems running at all times? With enough consecutive bad results, they might turn themselves off -- and just paper trade. When the results begin indicating that the market is favoring a given strategy, that strategy would begin trading again, noticing that expected return has gotten to its statistically significant threshold.

  8. Corey


    Furthermore, how many descriptors do you guys feel are necessary to describe the market condition?

    Is it a case of 'the more specific, the better'? Certainly, if we could identify the market condition with great accuracy, we could employ niche trading systems that exploit the condition.

    On the other hand, it may be best to utilize only a few, simpler descriptors.

    Volatile, non-volatile, trending, noisy, mania, panic, et cetera...
  9. Traders who have been around awhile know that no one system works in all markets, so they write systems to exploit a variety of markets, as you suggested. If there is a single 'black box' to work all the time, it is either held by the wealthy in secret or otherwise is sold for millions per subscriber and would not be available to most ET posters.

    Kudos for recognizing that a trader must be able to write their own systems, instead of relying on gurus or other's systems.
  10. tommaso


    Why don't you begin to put down some ideas.

    I would have a few ideas about the first 3 you mention, but I am especially curious about the "mania" and "panic" !

    I think it would be fun to discuss those. What do you mean by "mania" ?


    G-Bot - Automated Day Trading System Project
    #10     Oct 28, 2008