Not logged in
Viewing 8 posts - 1 through 8 (of 8 total)
  • #237902|
    Customer
    5 Posts

    I can get some fantastic results from SQ3 however after running tests on different time frame, different markets, slippage, profitable years beyond in & out of sample, robustness tests and wfm tests i think ive found profitable strategies but i then load them into my mt4 strategy tester only to find the time has been wasted with a flat equity curve or acct blown. I have very few strategies which actually show good results. im wondering if anyone knows what im missing or does everyone have the same problem?

    #238009
    tomas262
    Administrator
    1133 Posts

    Hello,

    we had some issues with SQ3 / MT4 results match for certain trading logics. I suggest you to update to StrategyQuant X version and try this solution. Follow https://strategyquant.com/download/ to obtain the latest version

    #238019
    Customer
    76 Posts

    Ups! I have a bunch of SQ3 strategies running live some of them brand new. Are these trading logics you mentioned identified somehow?

    #238023
    Customer
    62 Posts

    I had very similar issue also with SQ-X version. I generated many stable strategies, but after exporting them to MT4/Mt5 the results where no even close to that presented in SQ-X (re)tester. The issue lies in tick/M1 data. You need to make sure you are using proper data for strategy generation. The options are:

    – use Dukaskopy tick/M1 data and also Dukascopy as your MT4 broker (you need to clone downloaded data and synchronize GMT shift and DST!)

    – Find a reliable broker providing long history of M1 OHLC data, export it from MT4 and import to SQ-X and use for generation

     

    I hope it helps.

    Gr

    Chris

    #238029
    Customer
    18 Posts

    I find Dukascopy M1/Tick data is fine for MT4 backtesting and should give a reasonable backtest comparison when using the same data to build the strategies, your own broker data is of course preferable if you can get enough M1/Tick history to be viable, however you will probably only get a few months of M1 broker data from MT4 at best and whatever you do never download data from the MT4 history center, this is provided by Metaquotes not your broker.

    As coensio also mentioned you must always use the same GMT offset and DST settings as your broker in MT4 and StrategyQuant, this is very important otherwise the results will be way off when you move to a demo account or live trading and could lead to unexpected results

    #238504
    Customer
    7 Posts

    Hi Ash24FX,

    I was wondering what you consider to be “reasonable backtest comparison” between MT4 and SQ. I’m still trying to wrap my mind around what I think I a reasonable difference, or what is too much.

    #238506
    Customer
    1 Posts

    I take the approach of doing many tests/timeframes and so forth. I pick the ones with the highest profits/Sharpe>1 and Fitness>35.

    I then convert to MT4 and upload. Then I test each and yes – many high-profit trades in SQX bomb in MT4 testing. BUT – about 15-20% show great backtesting.

    I enter those into a spreadsheet and sort by MT4 profits. For speed I use 2 years worth of data in SQX and 13 months in MT4. So the profit numbers won’t be the same, but I go with what MT4 is giving me. Then I take the top 4-5 and trade live with 0.1 lot.

    I just started this on SQX and so far I have had some profitable trades. I am going live with a fresh batch of selected strategies this week.

    Then, as I have before I discovered SQX – I take stats for the week and see how it went. I use the comment field to tag the trades an have home-grown software to analyze what works and what doesn’t.

    So then, after a couple of weeks or more of live stats – I take out the non-profit ones and replace with fresh ones. The ones showing good profit after say 20-30 trades, I up the lot.

    This has been my go-strategy for a long time using various home grown EAs.

    Now I will have to be patient and see live what the strategies do. It would appear right now that I might just trade SQX strategies… this is one heck of a program. But it takes lots of testing and after-SQX weeding out.

     

     

     

    #238540
    Customer
    7 Posts

    Thanks for the extensive answer Ingvar …

Viewing 8 posts - 1 through 8 (of 8 total)

You must be logged in to reply to this topic.