Backtest Time Frame - newbie question

Discussion in 'Trading' started by Willleung, Jun 19, 2006.

  1. Hi,

    I had been doing backtesting with 250, 500, 1000, 2000 bars over the few days.

    Which result should i trust?


  2. Could you perhaps state the duration of a bar? Unless, of course, you don't think that is relevant.
  3. 1 day.
  4. And what would be the maximum number of bars used by the system you are testing (e.g. MA, stochastic...)?
  5. Max on all the systems tested so far are 26 for MACD, all EMA or SMA are below 17.
  6. Umhmmm. And what is the market you want to trade (e.g., S&P, NASDAQ, commodities...)?
  7. Quah


    You should trust them all - afterall, they are what they are - results. Why wouldn't they be correct and trustworthy?
  8. Testing for Hong Kong HSI & HSCEI futures...
  9. I can locate zones for variables, but find the zone changes with different time horizon.

    Doing test with 4000, 2000, 1000, 500, 250

    Some of combinations of variables hold up really well, being in top 10% in all periods, is it good enough? (2 variables system)

    I have some slightly more complex systems in testings like 4 variables, the zones moved in time.

    Should a person trade with testing done for the resent market condition or should trust the longer time frame?
  10. What kind of Z-score or other measure of statistical confidence do you want for the results?
    #10     Jun 19, 2006