Would this System Work, is it too simple?

Discussion in 'Strategy Building' started by mahler, Jun 3, 2005.

  1.  
    #11     Jun 3, 2005
  2. mogul

    mogul

    I don't know the period your are referring to above (04/06/28-05/03/31) but yes this type of system has worked pretty well on the Russell 2000 (especially emini). Just look at the daily charts, very large and clean movements in both directions.

    But as Spike500 pointed out, to backtest using one set of optimized settings is dangerous.

    To get a better feel for why that is dangerous, backtest year by year from the beginning and pretend you are actually trading it each year. That means you would re-optimize let's say each year and run the new optimized numbers forward.
     
    #12     Jun 3, 2005
  3. That's what i meant in my previous posting but i couldn't explain it clearly enough in english.
    :confused:
     
    #13     Jun 3, 2005
  4. It's easy to test if your system is good:
    the system should generate the same return (more or less) if you change all the parameters within a reasonable range. So this would be exactly the opposite of what people are trying to do here.
    You have to filter out the noise and keep the essential. Optimizing parameters is equal to adapting your system to the noise, so just the opposite of what is needed.

    The explaination is that a system needs stability, and stability comes with variable parameters that don't influence your returns.

    If i reread my own posting it sounds like i know something about trading. You can't even see that i'm a newbie.


    :D
     
    #14     Jun 3, 2005
  5. tpk

    tpk

    it's usually better not to have the same signal for entry and exit.
    u might try a volume for exit.
     
    #15     Jun 3, 2005
  6. my settings are 1, 2, 3 :eek:
     
    #16     Jun 3, 2005
  7. Right, keep it simple, stupid. But you have to use a daily chart, or you get too many signals.
     
    #17     Jun 4, 2005
  8. It is so simple that 95% of the traders lose money.
    Many professionals have huge drawdowns, reputated traders wipe out complete funds.

    But indeed it is VERY SIMPLE.

    I 'm curious to see the track records of those who find it very simple. But that's probably also too simple.

    My preference goes to 69, at least i can let my imagination go and have a pleasant feeling.
     
    #18     Jun 4, 2005
  9. I did the test on the last week trading:
    I reduced the parameters of my basic system by 30%.
    I replayed the signals.
    I got exactly the same signals as with my default settings.

    I wasn't surprised at all because i always focused on building a stable system that would not generate different signals because of some bad ticks or a bizarre move. So each part of the system tries to keep the rest in balance. Therefore you need the different parts to approach the market on different data. If the different parts of the system are too related they will lose all their value. They will only confirm each other and they will always do because in fact these parts are different versions of the same approach.

    So to me it is clear that optimizing parameters is useless.
     
    #19     Jun 4, 2005