Announcement

Collapse
No announcement yet.

Partner 728x90

Collapse

Backtest completely unreliable

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

  • NinjaTrader_ChelseaB
    replied
    Hello firebird631,

    No, this does not. You would need to use exit orders and not set methods to implement 1-tick intra-bar granularity.

    The 1 tick series should be an added series.

    Please review the example script.

    Leave a comment:


  • firebird631
    replied
    When it is wrote :

    Set methods such as SetStopLoss, SetTrailStop, SetProfitTarget, are submitted to the primary series only.

    Neither intra-bar granularity nor order fill resolution will allow a 1 tick series to update or fill these orders.

    For the managed approach, use Exit orders such as ExitLongStopMarket() and ExitLongLimit() instead of set methods for order fill resolution to apply.
    Does that means if the main serie is 1 tick it would permit the tick precision for exit too ?

    Should I prefer to have 1 tick resolution as primary, and manage order from the primary and keep others for strategy ?

    Leave a comment:


  • NinjaTrader_ChelseaB
    replied
    Hello firebird631,

    The 1-tick series is used for order fill pricing information which improves the accuracy.

    Below I am providing a link to the support article linked in post # 2 on 1-tick intra-bar granularity.


    Leave a comment:


  • firebird631
    replied
    I have added 200 tick series or seconds series, not a "1 tick". Effectively I managed order entry and stop update into the most precise tick serie.
    Then you suggere to add a 1 tick serie, but I could not manage correctly the indicator for that.

    Leave a comment:


  • NinjaTrader_ChelseaB
    replied
    Hello firebird631,

    May I confirm you have added a 1 tick series with AddDataSeries() and the order methods are being supplied the BarsInProgress index of the 1 tick series?

    Please provide the output text files where the data and order information was written from both real-time and backtest (as demonstrated in the RealtimeReplayHistoricalComparisonsTest example).

    Leave a comment:


  • firebird631
    replied
    Same problems. Trying with or without tick replay, increasing precision, adding data series ... all the time my strategies make 100000$ or more per month with 1 mini. Wow Trying the same in playback and it is not the same sing. The strategy sometime win, sometimes loss... more or less the same results in live. The strategy analyser is only a mascarade. Never use it butr try some serious plateforms for backtesting.
    Even in live. I have set the exact same strategy, same parameters, started the same time, on two sim account. From the third trade there is a difference, one execute, the other not, and take the signal 5 minutes later making a loss. End of the day the results are not the same. That's crazy.
    Lot of issue, lag, imprecision, occults behaviors, bad performance, freeze, memory leaks.

    The most fun, using Optimizer tool make totally reboot my i7

    I used lot of time my own trading robot with a genetic algorithm over the many cores of my CPU, during millions of test, on Linux, with perfect stability. I don't know how NT8 is made but it is not what I would pay for.
    Last edited by firebird631; 12-14-2024, 04:15 PM.

    Leave a comment:


  • NinjaTrader_ChelseaB
    replied
    Hello samish18,

    To confirm you have implemented 1-tick intra-bar granularity and enabled TickReplay while testing over the exact same datetime range, and the output text files from real-time and backtest are showing different condition values?

    May I have the output text files from the real-time and backtest?

    I may be able to assist with spotting what is causing the differences.

    Leave a comment:


  • samish18
    replied
    Please look at my previous posts, I have gone through all the links you have sent. This issue is apparent with every strategy and to a large degree (taking different number of trades per day, thousands of dollars difference). I and many other people pay to use this platform, if you are unable to provide solutions, please refer me to someone who may

    Leave a comment:


  • NinjaTrader_ChelseaB
    replied
    Hello samish18,

    You can expect that a strategy running real-time (live brokerage account, live market simulation, Market Replay etc...) may produce different results than the performance results generated during a backtest depending on how much resolution the script is running with. Further, this difference may be more easily seen on certain Bars types (e.g. Point and Figure, Renko) than others due to their inherent nature in bar formation.

    Below is a link to the Help Guide on 'Discrepancies: Real-Time vs Backtest'.


    Adding intra-bar granularity can help with this.

    Intra-bar granularity adds a second data series such as a 1 tick series so that the strategy has finer granularity in the historical data in between the OHLC of the primary series. By using the BarsInProgress index of the 1 tick series as the barsInProgressIndex parameter for each order method call, this allows for more accurate trades by supplying the correct price at the correct time for the order to fill with.

    Below is a link to a support article that details adding 1-tick intra-bar granularity and includes a link to the official reference sample demonstration.


    Further, below is a link to a support article that discusses what causes differences, how to use prints or writing to file to understand what is specifically causing differences in your script, and how to mitigate these differences.
    https://support.ninjatrader.com/s/ar...rategy-Results


    Market Replay data are very large files. From the NinjaTrader servers you must download 1 day at a time.

    If this does not provide the market replay data desired you may want to look into a 3rd party that provides market replay data for the NinjaTrader platform.
    Click here for a google search link for 'Download NinjaTrader Market Replay Data'

    Leave a comment:


  • samish18
    started a topic Backtest completely unreliable

    Backtest completely unreliable

    The backtest is completely unreliable. Running live vs backtesting over the same period yields wildly different results for every strategy I try. I have asked for clarity multiple times on here to no avail. How may I download months of market replay data at a time to get a more accurate backtest?

Latest Posts

Collapse

Topics Statistics Last Post
Started by davider97, Yesterday, 09:42 AM
12 responses
66 views
0 likes
Last Post NinjaTrader_Jesse  
Started by XanderT, Today, 03:47 AM
9 responses
29 views
0 likes
Last Post XanderT
by XanderT
 
Started by totalnewbie, Today, 10:03 AM
3 responses
6 views
0 likes
Last Post NinjaTrader_Jesse  
Started by sgordet, Today, 10:34 AM
1 response
3 views
0 likes
Last Post NinjaTrader_ChristopherJ  
Started by DanielRaz, 05-06-2024, 12:54 PM
3 responses
47 views
0 likes
Last Post NinjaTrader_ChelseaB  
Working...
X