For those experienced in system development

Discussion in 'Strategy Building' started by prox, May 22, 2003.

  1. prox


    I have about 4 contracts worth of backtest results for a fairly simple volatility based ORB system. Basically it enters based on a calculated factor, uses a protective stop at the LOD at the time of entry for a long, or HOD for a short, and then exits MOC .. so it's fairly crude and could easily be improved with BE stops or expected range projection profit targets.

    I don't really know what to make of it, other than the basics and would welcome any feedback.

    This has also been backtested with the same month contracts on the NQ and YM with very similar results. Around 55% wins and a 1.05 W/L ratio
  2. prox


  3. prox


  4. prox


  5. nitro


    I can't read any of them.

  6. hi prox, im a trader and system developer myself, so looking at the 1.05 W/L ratio doesnt look good. you could probably yield better results if you can set up your profit target a bit better, as you have said.

    (is the W/L a profit factor, or is 1.05 just reflecting the 55% win rate?)
  7. prox


    Anyone else having problems reading the files ?

    I presume the 1.05 WL is the Avg. Win / Avg. Loss and if I'm not mistaken, that would also be the Profit Factor number ?

    Right, it was just tested as an idea and certainly endures pretty big losses when wrong. Some basic money management and BE moves should improve drawdown and a market based price projections may or may not improve average win.
  8. FinStat


    none of the files are readable :confused:
  9. eyecheck


    There are lots of checks that one has to do to verify a system but one of the simplest rules of thumb that I have heard seems to work well in keeping me out of trouble. I use it as a quick screen to see if I am even remotely interested.

    Assume your actual win rate will be 10% worse than backtest. If you can still break even, system is probably worth exploring further.

    So at a 45% win rate, you would need a 1.22 avg win/avg loss ratio.

    With the numbers that you list, there is a good possibility that being postive is due more to luck than a good system.

  10. Hi Prox, as you eluded to if losses can be controlled this could be a viable method. Breakouts, at least in my testing, perform considerably well over the long term. I would highly recommend testing over longer data files though. If this is a dt system then you could even just combine your contracts.
    #10     May 22, 2003