How would you challenge this algo?

Discussion in 'Automated Trading' started by Drawdown Addict, Aug 5, 2023.

  1. themickey

    themickey

    Distance x time is all that's reqd.
    There are fancier excel functions; LINEST, LOGEST, TREND, INTERCEPT, SLOPE, FORECAST, GROWTH.
     
    #11     Aug 5, 2023
  2. virtusa

    virtusa

    That would be a real improvement, but it is not the holy grail. In strong trends you can have retracements after which the trend continues. These retracements can go thru the angle, kicking you out and missing the next directional move .
     
    #12     Aug 5, 2023
    themickey likes this.
  3. Time has sovereign control over profitability, so stratify your data according to the five days of the week and ask the question how is the profitability on Mondays? Tuesdays? I usually go back 10 weeks and average the profitability for each of the five days. Then within each day stratify the data further for longs versus shorts, again looking for strengths versus weakness. You may find some days consistently make money and others consistently lose it, while longs may work well on one day but shorts the other. And that way you can trade towards strength and away from weakness.
     
    #13     Aug 5, 2023
    rb7 and Drawdown Addict like this.
  4. Profitability per day is shown in the bar chart. It is the delta between short and longs, so I will need to separate that in two charts. I can do that, no worries.
     
    #14     Aug 5, 2023
  5. rb7

    rb7

    6 months of backtesting results is not enough. You need at least 2 years. And results from out-of-the-sample data as well.
    But it seems promising at first
    Profitable algos don't need to be complicated. Sometimes, simple rules with a few tweaks all that you need.
     
    #15     Aug 5, 2023
  6. MarkBrown

    MarkBrown

    this is a juggling act in the mind to perform.

    i would determine the 75 percentile reversionary price action and define that with boundaries.

    anything outside of that would be a trend which could be ranked according to how far it deviates from the mean 75 percentile movements.
     
    #16     Aug 5, 2023
  7. I am going to try those percentiles for sure.

    Also I should try something like this article:
    https://trendspider.com/learning-center/linear-regression-slope-a-comprehensive-guide-for-traders/
     
    #17     Aug 5, 2023
    NASRI likes this.
  8. Good Morning Drawdown Addict,

    Please do a back test on strategy from January 1, 2006 to August, 4 2023. View the report.

    When you see the results, EVERY QUESTION you have will be answered in a matter of 1 minute. You will KNOW exactly what to do next. EXACTLY what to do.

    Always remember, if you want Guranteed Rest, you must Backtest.

    Thank you
     
    #18     Aug 5, 2023
  9. back testing in train station is reliable, back testing and ninja is a waste and corrupt
    https://ninjatrader.com/support/helpGuides/nt8/?discrepancies_real-time_vs_bac.htm
     
    #19     Aug 6, 2023
    SimpleMeLike and rb7 like this.
  10. #20     Aug 6, 2023