Are the results of backtesting reasonable?

Discussion in 'Strategy Development' started by yoyo2000, Jul 4, 2005.

  1. yoyo2000

    yoyo2000

    This topic is initially from Trader's Roundtable forums,and have puzzled me for a long time,any suggestion is appreciated.

    Several weeks ago, when I test my system on a basket of stocks,I met a problem which puzzles me till now:

    The 2nd Pic is the result of that system performed on 100 stocks out of a whole market,slippage and commission are taken into account,100 shares are hold in each trade.It's not an acceptable one,but when I changed the rules of money management into 30% of account cash for each trade,the final result is completely different(in the 1st Pic,all the other trading rules are of the same.)

    The direct reason is,for the latter rules,many of the trades are wiped out because of the lefte cash:after several trades,the left cash is not enough for the coming position,so the chances of trades are ignored,till a positon is released.
    And this situation is met frequently for basket trading,which performence result should I believe?
    In the case of actively timing the stocks,should I reduce the number of the stocks in the basket,or do something others?

    When a system bases on a basket of stocks,commodities,or currencies,it possibily meets the problem of the limitation of account cash or other fixed position rules(for example,max 10% account capital for risk,or max 10 position,ect.).They therotically wipe part of the trades,and maybe,the characteristics of the system are changed.

    Any suggestion for it,please?

    Thanks in advance.
     
    • 1.jpg
      File size:
      41.7 KB
      Views:
      97
  2. yoyo2000

    yoyo2000

    this is the 1st pic.
     
    • 2.jpg
      File size:
      41.5 KB
      Views:
      77
  3. How far back are you going with this backtesting, anyway?

    Are you using daily or minute data? And which platform are you using?

    How about a pic of the equity curve?