PRT Automated trading - issues and suggestions

[Guest Stef]

Hi guys,

 

Some issues that I have picked up with PRT's automated trading module (ProOrder), and some suggestions:

 

* Probably the biggest issue is that the backtesting (Probacktest) and automated trading modules (ProOrder) do not behave the same. So, if you get great results with your backtest, don't get too excited, as the automation of this will most likely produce vastly different results. Some of the differences include:

 

-Pro order will trigger on the open of the next candle after the condition has been met but Pro back test will register the trigger of the trade on the close of the same candle the condition was met.

 

- Pro Back test cannot determine will your stop or limit will be triggered first if both would've been triggered in a specific candle. In this situation Pro back test will show the trade as if the Limit was triggered first. There is no setting yet to change which of the stop or limit will take priority when this happens but Pro real time is doing development to improve this feature.

 

- Pro back test will calculate the results using a step of 1 for the trailing stop and this value cannot be changed when testing. Pro Order will execute the step of the trailing stop by using the minimum allowed step distance of the particular market. To determine what the minimum step distance for a trailing stop on a market would be you can open a deal ticket on the IG platform and tick the trailing stop box and the step field will appear indicating the minimum value.

 

* ProOrder allows you to view the performance of your system while it is executing. This is great! Especially seeing as Probacktest behaves differently. But, there are some problems and shortcomings:

 

- You can add indicators to the price section of the chart, but not as a separate section underneath!? This is great for displaying PRT or custom indicators that you use as part of your strategy that can be plotted on the price section, but not so great for others that are normally displayed underneath (e.g. MACD, RSI, etc).

 

- Any indicators that you add will not be there the next time you open the window to display performace again! Common guys! What is the point of being able to add these indicators in the first place? This is especially problematic if you have your own indicators that need a lot of changes to its display properties (line styles, width, etc). It takes a huge amount of time to redo this every time for every system (if ProOrder and ProBacktest behaved the same, this would not have been that necessary, or that big an issue).

 

- Resizing! The view performance window allows you to resize the sections. But, again, if you reopen it, the sections are back to their original sizes. Why do the equity curve and positions sections have to take up so much space? I am more interested in the price section. This same issue applies to Probacktest also.

 

That's it for now!

 

Regards

Stef

[Guest TimP]

Hi Stef,

 

Thank you very much for taking time to give us feedback. I have passed this onto ProRealTime for you and I do have to thank you for the detail you have gone into with your feedback.

 

As soon as I hear form the PRT team I will let you know and as always if there is anything else you would like to add please let me know. 

 

Thanks,

 

Tim 

[Guest TimP]

Hi Stef,

 

Just to let you know that I have just been in contact with ProRealTime regarding your feedback. They would like to thank you for the thought you have put into your feedback as such clarity really does give them a lot to work with and I hope one day to see a platform that has followed the legacy of the points you have made.

 

Thanks again! 

 

 

[Guest ws01]

Yes this is so true, especially if you are working on 5 minute charts or longer. I have backtested fantastic queries only to find that absolutely NONE came out as expected..

Even worse when you then back test what you ran live, it shows great results still, even though it actually lost horribly. What a learning curve I have been through and this should be made clear in the Manual.

Its not enough to say that backtested results might not match reality. In fact it is impossible to trust the backtest because of the issue refered to by Stef. Just imagine that you have an automated query thats based upon a 1hr chart.. In reality you might be stopped out after 1 minute and yet backtesting will wait for 1 hour, and the price is very different in most cases....

Even working with 1 min charts still produces very differing results...