No Support From ProRealTime

Forums ProRealTime English forum ProRealTime platform support No Support From ProRealTime

Viewing 15 posts - 16 through 30 (of 34 total)
  • #100024

    The fact is that the new backtest engine is more accurate than the older one. This is that engine that will be released in June, incorporating MTF capabilities for every account, as you know.

    For your personal case, I think that you should wait for the complete analysis and the feedback from support (remember that when you are an IG customer, the time to get a return could take longer).

    #100034

    @Nicolas I do not believe the issue with the strategy I emailed to you is an isolated incident. Was it able to backtest successfully on your side? If not this means there is a greater issue as it used to work fine. I would really appreciate if you could spend 5min on it to see if you are experiencing the same problem. Others I have sent the strategy to are experiencing the same problem. I can also not understand why the backtest speed with larger MTF strategies have been reduced so drastically. After the update it easily takes more than 3 times longer, plus the results look way different to what they did before.

    #100069

    Hello Juanj,

    I face also issues with my codes, there should have been a regression somewhere, but don’t know if this is related to your problems.

    More specifically, I have strategies running live with a fixed target price. I’m not using a breakeven or a trailing stop, but remember having tested them in the past.

    Now that I want to reintroduce breakeven or trailing stops in my code, the backtests are running into an infinite loop. So to be clear, my strategies are fine isolated, the breakeven or trailing stop are fine when used with simpler codes, but together they can’t be tested..

     

    #100073

    It’s very possible that the issue is indeed related to the latest update. As mentioned I only experience problems with my more complex pieces of code, and I actually also remember getting an infinite loop error a couple of weeks back when trying to use a rather hefty loop in one of my code. I have actually decided to stop coding further on my MTF PRT codes until this ‘yet unidentified arithmetic’ issue is resolved as the backtest speed is unbearably slow and makes testing a pain.

    #100076

    I made some tests with your strategy juanj, none of the variables are calculated. There must be something in the code that make it crash before it starts.. I will try to know more about how it goes in the investigation today.

    1 user thanked author for this post.
    #100077

    Thank You Nicolas, I sincerely appreciate you looking into this! It’s almost as if the new engine just gives up and stops trying to calculate the variables.

    Like I said it literally worked fine for months and then all of a sudden stopped working. And at the same time, I noticed the increased backtest periods as well as the strange arithmetic calculation anomalies.

    #100082

    As you know, the new backtest engine (that embed the MTF capabilities) is in beta testing and was only available for people asking its activation, and with a warning message. So it is still possible that some issues (and differences!) regarding the old and new version behavior could emerge.

    #100084

    Completely understandable and expected but obviously the same understanding exists that these issues will be resolved in a timely fashion as they are brought to light. Which is obviously the reason why we are pointing them out.

    #100176

    Ok make sense that there is an issue with the new backtest engine, because I’m also activated for MTF

    #100574

    @Nicolas with regards to “I made some tests with your strategy juanj, none of the variables are calculated. There must be something in the code that make it crash before it starts.. I will try to know more about how it goes in the investigation today.”

    Is there perhaps any feedback from your side on the investigation?

    #100576

    Yes, I had feedbacks yesterday, still investigating..

    #100641

    Looking forward to the outcome and resulting fix.

    #101163

    @Nicolas after nearly 2 painfull months of waiting it would seem the above mentioned PRT bug is finally resolved! As of today, all strategies are running normally again, including the one you submitted for investigation. Would it perhaps be possible for you to find out more details as to what turned out to be the problem?

    #101268

    Internal error detection fix of the parsed code, new engine upgrades made on 20th of June, should have fixed your issue.

    MTF / new engine is coming and should be released completely this Wednesday.

     

    #101271

    An interesting question would be whether PRT has the necessary QA and internal checks and balances in place to find this type of error without the community reporting it?

Viewing 15 posts - 16 through 30 (of 34 total)

Create your free account now and post your request to benefit from the help of the community
Register or Login