Evaluation of Backtested System Results

Discussion in 'Strategy Building' started by trend456, Apr 23, 2003.

  1. This has nothing to do with evaluation but some interesting things happening to my systems that I don't appreciate.

    My systems are working fine but I'm getting more size as the system performs better. It's nice to trade more size but adding size to a system is different from adding a size to a discretionary trader. It makes the system look more volatile when the size is added in a wrong way especially when size is added too early.

    Everytime the size increases it's seems like I'm starting back to the beginning.

    Seems like I'm back to the start again.......:(

    Good Trade!

    Trend:)
     
    #41     May 1, 2003
  2. Thanks.

    I see a lot of this at a prop. Especially with the youngsters. There favorite quote,"I will make millions in a year!"

    LOL

    Trend
     
    #42     May 1, 2003
  3. Just wanted to add another quote,"We can add 500 lots and we'll be rich!"

    LOL

    Trend
     
    #43     May 1, 2003
  4. Why do you look at win %/loss size/ win size as separate distributions, rather than treating all your trades as a single sample of both positive and negative values?

    -bbc
     
    #44     Oct 3, 2003
  5. olintner

    olintner

    A lot has been said about finding a robust parameter set for a given trading system. Graphical interfaces like Excel facilitate that task - but how about robustness over a range of markets.

    Most systems profitable in the fixed income, foreign exchange and commodity world don't work on equity indices. Even if equities are left out - the one system for all markets approach still gives a good feel in terms robustness.

    Last but not least, even when trading longer term, the most important parameter in my point of view will always remain slippage - the cost of doing business. slippage even effects long term systems and defines the range of frequency tradeable.
     
    #45     Oct 10, 2003
  6. decemb

    decemb

    Comparing performance during the same test period but beginning from different dates can open your eyes on unexpected but really possible account crash - Results differ tremendously - see the report "Worst Case Analysis" http://www.traderstech.net/WCS.htm.
    This approach needed to realistically appreciate MAX DD and other parameters before using your beloved systems.
    Be aware! Good luck!


    :confused:
     
    #46     Oct 20, 2003
  7. olintner

    olintner

    Hi Decemb,

    tried the link but it didn't work - could you please check?

    Regards,

    Olli
     
    #47     Oct 20, 2003
  8. rickty

    rickty

    #48     Oct 20, 2003
  9. olintner

    olintner

    #49     Oct 21, 2003