Difference between ProOrder Auto Trading and Live Backtest

Forums ProRealTime English forum ProOrder support Difference between ProOrder Auto Trading and Live Backtest

Viewing 15 posts - 1 through 15 (of 15 total)
  • #122077
    VN

    Hi,

    I am testing a strategy and I am comparing how it performs by running it through the ProOrder Auto Trading method (executing through my IG demo account) and at the same time running it via my PRO Backtest strategy of my demo ProOrder application. I am then comparing the orders as they happen in real time, looking at the opening and closing times between the two methods and most of the time I am seeing the same results (as one would expect since they are the same strategy) but not always!

    For example, today I noticed that an order to go short was opened and executed on my IG demo acc via the AutoTrading method, but at the same time that same order was not made on the POR backtest method which is running simultaneously to the AutoTrading tool.

    Another example of a difference was again earlier in the day, I noticed that an order was opened at 12:54 on the PRO backtest method and closed at 13:09. However, the same order via the Auto Trading method was opened at 13:03 and closed at 13:09 to match the backtest method. On this particular example there appears to be a difference of 9 minutes in opening the trade between AutoTrading and Backtest methods..

    Is the market data and the timing of that market data the same between AutoTrading and Backtest modes the same?

    Has anyone come across this behaviour and if so do you know why these differences exist?? Any insight would be greatly appreciated!

    #122078

    Is the market data and the timing of that market data the same between AutoTrading and Backtest modes the same?

    Backtest has fixed spreads, no slippage and no variable minimum stop distance.

    Forward Test has variable spread, variable slippage and variable minimum stop distance.

    1 user thanked author for this post.
    #122080
    VN

    Hi Grahal,

    “Forward Test” in your reply, is that the same as AutoTrading mode?

    Just want to clarify that I am obviously looking at the backtest results but that is in REAL TIME, meaning I am looking at the signals and orders as they happen now and comparing them to what the AutoTrading mode is sending to IG. I am not comparing orders in the past like yesterday or last month but orders as they come in.

    My orders also happen at the close of each candle (1 min as example). It feels like there is some difference in those candle closes between the 2 modes otherwise the orders and their timings must be identical..

    #122102

    “Forward Test” in your reply, is that the same as AutoTrading mode?

    Yes above is using Real Live data with spread, slippage, min distance at the whim of the broker.

    orders as they happen (using the backtesting window).

    Above is using Live data but with spread fixed (at what you set it in the Platform) no slippage and no broker imposed minimum stop distance (only what you code in to your strategy to simulate what the broker might decide to do! )

    #122148
    VN

    Thanks for your response Grahal

    can you clarify what is “Forward test”? Are you referring to running a strategy on market data as it happens through the “pro backtest”?

    #122150

    Yeah it’s all  bit confusing.

    Forward Test is when you set a System going using the window attached.

    Forward Test can be either on the Live Platform (real-time trading, real money) or Demo Platform (paper / virtual money) and results should be the same.

    You may see use of the term Real Live (real money) or Real Demo (paper money) … this means using Real-time data … on a Forward Test.

    #122153
    VN

    ok, so using your terminology I am comparing the following:

    Forward test on Demo platform. let’s label the resulting trade orders from this test be “list A”.

    Running the same strategy on backtest mode, but only looking at the trade orders that come in after the latest time has expired ie if the time I kick of the backtest is say 16:00, I will be looking at any orders that may get executed by the strategy running after 16:00. Basically looking at the orders as the price changes after I have ran the backtest and just leave the strategy running. Call this list “B”.

    What I am seeing is differences between list A and list B.

    does this makes sense?

    #122164

    Yes and most likely it’s fault of the code. When there’s no market position, when the backtest triggers, so should the running system on demo/live account. I’am not sure when criteria is hit and the next bar is empty where it should buy at the open.

    #122166

    I see List A and List B as both on Forward Test (FT) (trades after 16:00 on ListB).

    I stand to be corrected?? 🙂

    To be clear … we need to use PRT window specific terms?

    1.List A is on FT in the ProOrder window. (True Forward Test)
    Variable spread, variable min stop distance, slippage … all 3 set by broker.

    2.List B is on FT (after 16:00) in the ProBacktest window. (False Forward Test).
    Fixed spread set in Platform, no stop distance unless coded in, no slippage.

     

    #122171
    VN

    ok so I understand you so far.

    For list A, is there a way I can visually plot (graph) the variables which my strategy uses to determine the entry/exit conditions that it had when the orders were generated?

    For list B, this is very straight forward as you know with the backtest window you can just graph any variable using the code editor even if the orders took place after 16:00 as it would plot any variable up to the point where there is price close available. With the Autotrading mode however (the Forward testing on demo using your terminology) I can only see the strategy performance and trades executed. I cannot plot variables and/or indicators retrospectively. What I want to see is the price history on which list A (forward test demo) was generated and compare it to the price history of the backtest. I should then be able to see differences in the candles e.g. different opens and closes given what you said about the variable spread.

    (hopefully you understand me 🙂 )

    #122177

    Yes I understand and the answer is … NO! 🙂

    I have often thought it would be very useful to see just what you want to see, I’m sure others would also.

    But no … there is no way to  see variable values / crossing etc under True Forward Test (True FT) conditions.

    #122179
    VN

    damn, I was hoping you would say “YES”, lol..

    Anyway, I was thinking about what you mentioned in your reply yesterday (122150), “Forward Test can be either on the Live Platform (real-time trading, real money) or Demo Platform (paper / virtual money) and results should be the same.”

    If the results are the same then, don’t I just need to load the live PRT chart as those opens/closes will be the actual opens/closes on which my list A trades would have been running against and then compare it to the Backtest price curve?

    #122180

    If only it could be that easy!

    When I said they should be the same I meant should as in desirable to be the same.  Sometimes they are, but often they are not exactly the same. If market volatlility is low and you have set the spread same as in Real Live then you have a very good chance they will be the same.

    If volatility is high then broker may increase spread and minimum distance and of course there is more chance of slippage with high volatility.

    #122182
    VN

    ok useful to know this Grahal.

    I use IG as broker and I have noticed that they do widen spreads accordingly as the underlying market vol changes drastically as it has been over last few weeks. That being said however whatever the spread is at any point in time, I would expect it to be reflected to PRT live instantaneously as people can trade of PRT directly so it cant be that IG live has one price and PRT live has another.

    However if I understand you correctly you are saying that under high vol times the Forward test DEMO prices may not get updated with the latest spreads at the same time as PRT live. Is that correct? If demo doesn’t get updated at the same time however as PRT live, then why would there be difference between FT Demo and Backtest as backtest prices you mentioned are with fixed spread etc..

    #122184

    Forward test DEMO prices may not get updated

    As we are getting right down to detail now I think it better to use the terms True FT (Real Live data) and False FT (Backtest data).

    Also I thought all the discussion is about running on PRT Platfom.

    I would expect it to be reflected to PRT live instantaneously as people can trade of PRT directly so it cant be that IG live has one price and PRT live has another.

    I would expect also else we could arbitrage between the two Systems (as if we would have fast enough connections! 🙂 )

Viewing 15 posts - 1 through 15 (of 15 total)

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