I have been backtesting on my live and demo account using PRT on IG and the results vary drastically is there any known issues with differing data feeds for each?
The data should be identical from IG for demo and live.
There can be a number of reasons for differences from different spread used in demo compared to live back test or even strategy specific reasons such as using a weird time frame (26 seconds for example) and then not starting both tests from exactly the same time or even something silly like having tick by tick data selected in one test but not the other.
You might want to post some screen shots of each test to show the differences and maybe post the code that you are back testing so we have all the info to work on and test for ourselves.
Hi David,
All the Demo accounts are using a new engine (multi timeframe) which has a different time execution than the old engine.
The Live accounts are still on the old engine however you can request the activation of the Multi Timeframe on your Live account upon request.
Please send an email directly to IG with your request and they will do the rest.
Have a great evening!
Experiencing the same thing. Not even close inn terms of results and I’m not using different timeframe references in my code. It’s very simple code.
But I have noticed small discrepansies for a while but this one had me falling of the chair.
60M scalping algo.
First image is live account and second is demo.
@Jesper Time to analyze some of the trades to know where are the differences! Why orders are triggered and why do they close, do they respect the strategy or not? …
@nicolas I will look in more closely on some of the trades.
@nicolas The real account trades to the left and the demo account trades to the right. They are taken at pretty much the same place but stops are triggered in the demo environment.
I’m thinking this is a fault in the live account engine and not the other way around. My guess is that the new engine has more “in-bar” data and thus more representative of the real deal.
Does that sound possible?
This is a typical behavior of orders opened and closed in the same bar. You should look at inferior timeframe than the 1 hour one and verify when the ordres are closed, do they triggered really in the bar and at the good time?
My guess is that the new engine has more “in-bar” data and thus more representative of the real deal.
Definitely.
@nicolas I can’t tell when the orders are triggered since only bar-open time shows up in the list.
I’m content with the answer and most likely the reason why things are not looking the same.
Back to the drawing board.
I can’t tell when the orders are triggered since only bar-open time shows up in the list.
You can see exactly when the Trades Open and Close after a Backtest from the Detailed Report … see attached.
Ooops just noticed you may gave been talking about zero bar / same bar trades?
But if you run the backtest on the 5 min bar (not 1 hour which gives same bar trades?) then you probably will get a Detailed Report similar to attached with Open and Close times.
I would also like to give my two cents here.
I had similar experience in the past – well more than 1 year ago – and I explained it to me that on the Demo account we have a bit a different market information setup.
For example: the minimum lot size you can buy / short a financial instruments is not always equal as it is configured in live account.
I also think that the spreads are not same as demo and live.
Last but not least, who is the counterparty in demo? IG I assume with unknow size on bid/ask.
On live we have IG as market maker and of course client orders as well.
That all has to be consired before putting a well running demo strategy live.
But are there some concepts to follow if you want to make sure that you get approximately the same results in Live as in Demo?
