Announcement

Collapse
No announcement yet.

Partner 728x90

Collapse

This can't be right. Help me understand Walk Forward

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    This can't be right. Help me understand Walk Forward

    I'm trying to wrap my head around the walk forward portion of NT.

    I must not be understanding this correctly. The strategy I'm currently testing is performing around a 13 profit factor and 93% percent profitable. This just seems impossible. (Even in back testing, where you could say there was curve fitting, I was only getting a 3-5 profit factor with 70-80% profitable.) With walk forward, I've been picking both random dates and long consecutive runs with those same 90+% results.

    This is a 1min strategy. So, I'm using small optimization periods. My optimization period is at 2 days and my test period is for 1 day.

    Just so I'm clear. I have some questions.


    1) Using a 2 day opt and 1 day test literally means that NT will optimize for two days and use the best settings on the third day? It is not doing any curve fitting or optimizing on that third day? Is that correct?

    2) If I want to match these results (not taking into account slippage and everything else that goes wrong during live trading), I simply optimize every night for the last two days and use those results in my strategy for the next day of trading?

    3) I'm using NT 7, and the optimization settings are "Default 24/7". Since I'm running this strat on ES, when would be the best time to run my 2-day optimization to be ready for the third day of live trading? I hope this makes sense. I'm trying to figure out at what time NT would be running the 2-day walk forward optimization process (when using the Default 24/7 settings).

    Many thanks

    #2
    lookOutBelow, I would suggest then you run your strategy in Market Replay against the simulator to 'battle test' it after going through the backtest and walk forward.

    Your understanding is correct, optimize on the insample data (your 2 day window) and then apply to the out of sample test period (your 1 day) to evaluate how good it held up on those parameters.

    It would be run after the session ends, so midnight in the defined timezone.
    BertrandNinjaTrader Customer Service

    Comment

    Latest Posts

    Collapse

    Topics Statistics Last Post
    Started by helpwanted, Today, 03:06 AM
    1 response
    7 views
    0 likes
    Last Post sarafuenonly123  
    Started by Brevo, Today, 01:45 AM
    0 responses
    7 views
    0 likes
    Last Post Brevo
    by Brevo
     
    Started by aussugardefender, Today, 01:07 AM
    0 responses
    5 views
    0 likes
    Last Post aussugardefender  
    Started by pvincent, 06-23-2022, 12:53 PM
    14 responses
    242 views
    0 likes
    Last Post Nyman
    by Nyman
     
    Started by TraderG23, 12-08-2023, 07:56 AM
    9 responses
    385 views
    1 like
    Last Post Gavini
    by Gavini
     
    Working...
    X