Not logged in
Viewing 5 posts - 1 through 5 (of 5 total)

Forums>StrategyQuant>General Discussion>How much difference do people get between MT StrategyTester and Live?

  • #256814 |
    Customer
    4 Posts

    Hello all,

    I am wondering, how much of a difference are people getting on their live account vs mt4 strategytester? meaning how much of their live trading history actually matches the mt4 strategytester?  let’s imagine is the “perfect” expert adviser being tested/used in live. Is there such thing as the mt4 strategytester matching 90% live trading history?

    Do mt5 experts (as long as their properly “built” and pass all tests) offer a better chance than mt4 experts?

    #256862
    tomas262
    Administrator
    1821 Posts

    Hello,

    this is one of most frequent question around. This happen quite often actually during testing. We are elaborating on this topic in this article: https://strategyquant.com/blog/real-trading-compare-live-strategy-results-backtest/

    There can be many reasons for differences though. First, you could try to perform a high precision backtest in MT4 using Dukascopy data (for more info check this https://strategyquant.com/doc/quantdatamanager/test-strategy-metatrader-4-tick-precision/). If backtest results are similar then there might be a problem with your broker’s data meaning it differs too much from the data used for the strategy development. Another cause could be when a different timezone is used, there is a difference spread, slippage etc. WIthout providing all details possible it is hard to know the reason. It’s kind of detective work

    #256864
    Customer
    4 Posts

    Tomas,

    Thanks for your quick reply.

    I am glad I am not the only individual with this issue and I will take a look at the article you posted.

    I do understand that live trading will of course not be the same as the strategy tester but I was thinking it would at least be 50-80% (in my case around 35% of the strategy tester matches live trading).

    I have used high precision MT4 Dukascopy Data, even had a demo account with them to compare their strategy tester to my broker’s strategy tester (Oanda) and both results were always at least 90% the same.

    A little bit about how I searched for the strategy:  H1 XAUUSD Trend following (with some signals/indicators and stop/limit options I manually picked), order type was set to only “(Stop) Enter at limit” while under exit type I had everything selected. It always had stop loss and take profits (which according to some online search could be causing the issue). I will now search with no stop loss or take profit and only use indicators to get in and out of trades, and see if the strategy testers and live trading time entries/exits and prices vary.

    Here is the kicker, after doing all the tests in the re-test and optimizer I decided to save the file as a mt4 EA and dropped it in the  Dukascopy & Oanda MT4 strategy tester and the results on the strategy testers were almost twice as better than those that showed in StrategyQuantX (adding more to the confusion LOL)… maybe it could of been I was using excessive spread and slippage set in SQX? I had spread at 0.75 (since broker claimed they have 0.25 on average) and the slippage was 0.25-0.50.

    Could setting the search and live trading using MT5 be any better? according to more online research, MT5 is better for running expert advisers for those who know how to code (which is not much of an issue with StrategyQuantX).

     

    #256878
    Gianfranco
    Customer
    112 Posts

    I can tell you that with dukaskopy not ecn..the spreads you use by default of SQX data, plus 1/2 slippage are also good in real …. except for some rare cases of fast markets … sometimes slippage goes also in your favor

    with admiral market … slippage often high

    they are not opinions but real trades

    thanks gianfranco

    #256891
    Customer
    4 Posts

    I see what you mean… I am just not sure what could be causing trades in live that do not happen when running strategy tester in forward testing.

Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.