your criterion of the refusal the system

Discussion in 'Strategy Development' started by jenek-cowboy, Nov 27, 2007.

  1. When i must stop to traid my sistem?
    My opinion is that if my sistem has greater max DD that i see in Perfomance when i tested it then i stop trade this system and start when the equity line has breakout the maximum of equity.
    Have you other method?
    sorry for my english..thank you
  2. Fangdog


    The questions you must ask, Is it your system which is reason for failure or is it you which is causing the system to fail?
  3. i don't understand you.

    For example.
    We was profit among one year. Then it stoped give us profit.
    The question is:
    When we can say that the sistem failed?
  4. Stop your system when it shows drawdowns that are worse than 95% of the drawdowns that could have happened by chance alone.
    You can use for example a Monte Carlo simulation for that purpose: take your backtested results and run a Monte Carlo simulation of drawdowns.
    However, you must be sure that your backtested results are not overfitted, otherwise you will probably get the 95% drawdown .... in real time, during trading :)

  5. I can use this method in WLD. But why you take this number (95%) ?
  6. drawdown is totally normal, but it should be reduced as much as possible, even it this means you reduce your profits.

    if the system is no longer profitable, stop trading it with real money, analyze why the trades are failing that much all of a sudden, and adjust the parameters (bigger or smaller stop, extra filtering, etc)

    but i would stop trading it for now if you lost confidence.

    can you post some basic information on your system performance so we can assist further? like max dd, profit, win%, profit factor, max winner, max loser, etc
  7. 95% is widly used in statistics to define a confidence interval. Regarding the MC method for drawdowns, it just tells you that you should not encounter such a bad drawdown 95% of the time (for example 95% of the years if you evaluate drawdowns per year in your Monte Carlo Simulation).
    If it happens that you encounter a wider drawdown the first year, probably your system is not a good one (odds are 95% being correct in this conclusion).

    95% is arbitrary, you could use 99% for example but for 99%, the drawdown to wait for is probably much larger.
    You have to make a choice here : smaller drawdown/larger confidence to stop it for good reasons.

  8. yes ofcose.I will much interesting..

    I use 30 min timeframe.Test was on future on index RTS(it is based on index RTS.He consists the most large 30 companies.RUSSIA INDEX. Dayly sales are 1000.000.000 dollars per day). Two opt parameter for long and 4 for short. Costs are 0.05%
  9. you say that the system no longer pays? this wealth lab results screams profits. the drawdown is very acceptable.
  10. equity
    #10     Nov 28, 2007