EA Forex Academy – Algorithmic & Manual Trading Courses Forums Forex Trading Forum EA Studio Strategy shows very different results in MT4 Strategy Tester

Viewing 15 posts - 1 through 15 (of 15 total)
  • Author
    Posts
  • #13621
    Admin bar avatarSimon
    Participant

    Dear Petko,

    I have been watching a lot of your really great videos on YouTube, thank you very much for publishing them.

    I am contacting you because I am have been a user of Forex Strategy Builder Pro and EA Studio for many years now, but cannot get anywhere with it because I find it extremely difficult to get hold of and use accurate forex data.

    I was following along with your video – Forex historical data – how to export properly for EA Studio , and you show that you export new data from MT4, import it into EA Studio, generate one strategy purely for the purpose of making sure that you get the same results in the MT4 Strategy Tester as you do in the EA Studio strategy Journal, which you do successfully.

    I have followed the procedure as closely as possible, and the results I get in the MT4 Strategy Tester are nowhere near the same as the ones in the EA Studio Journal.

    Here are the 124 entries shown in the EA Studio Journal:

    Here are the 6 entries shown in the MT4 Strategy Tester:

    Here are the settings I set in  the MT4 Strategy Tester:

    I really hope you can help me figure this out, it’s stuff like this that keeps stopping me from progressing, I hit hurdles like this, can’t find any help, get distracted with running my other business, and then find myself back to ‘square one’ again.

    Please let me know if you need any more info from me, I will gladly find it out for you.

    If I can get the fundamental procedures organised for routine practice, I will no doubt happily buy several of your courses, but as it is, if I can’t get EA Studio or Forex Strategy Builder to work properly, I would be wasting my money.

    I hope to hear from you sometime.

    Many Thanks,
    Simon

    #13627

    Hello Simon,

    Glad to hear from you. You should be getting the same results with FSB Pro/EA Studio and MT Backtester. Before I give you some reasons, what you are missing, can you please post the following pictures:

    1. The strategy in EA Studio – with the entry/exit rules, everything, simply the Editor of it

    2. The MetaTrader where you do the backtest

    3. The statistics of your Data – go to EA Studio Data – Data statistics

    Kind regards,

    #13669
    Admin bar avatarSimon
    Participant

    Hello Petko,

     

    Thank you for such a quick response, please see images below:

    1. The strategy in EA Studio – with the entry/exit rules, everything, simply the Editor of it

    2. The MetaTrader where you do the backtest:

    This is a backtest on a live account – I get zero results/entries/exits on here, “Results” and “Graphs” tabs are empty – however, the current balance is only 36.05 GBP, would that be the reason?

    Settings Tab:

    Report Tab:

    Journal Tab:

     

    This is a backtest on a Demo account, the balance is closer to what was entered into the account settings in EA Studio prior to creating the strategy.

    Settings Tab

    Results Tab:

    Report Tab

     

    Journal Tab:

     

    3) The statistics of your Data – go to EA Studio Data – Data statistics

    Please let me know if there’s anything else you need.

    Kind Regards,

    Simon.

    #13672
    Admin bar avatarSimon
    Participant

    Sorry about the duplicate posts, The submit button didn’t seem to be working when I posted them, I can’t see a way of deleting them?.. at least one of them has the images.

    Thank you.

    #13722

    Hello Simon,

    Now here is what I see from the pictures:

    1. On your MT backtester, the journal shows an error. It should not be from the small money you have inside. You can set the trading amount to start the backtest. For sure it a big number. About that, you should connect your broker, and ask why a backtest could not be completed. Are you sure they allow algorithmic trading on this live server?

    2. The Historical data looks alright, but it is too small. Did you try to get more by pressing the Home key? If you look at your strategy, it has less then 100 count of trades which is really small.

    3. The spread is big! You should look for brokers with a smaller spread. With GBPNZD it should be about 2-3 pips. You have 7. Even you do everything right, wich such spread, you will hardly see profits into your account ever.

    I would suggest you give it a try with another broker with smaller spread and more Historical data. I am sure, this way you will do it all properly.

    Kind regards,

    #13764
    Admin bar avatarSimon
    Participant

    Hi Petko,

    Thanks for your reply, and for pointing out that the journal shows an error.. I looked up “mt4 ordersend error 131” and found out the following info:

    Handling OrderSend Error 131 in MetaTrader 4

    OrderSend Error 131 is a very popular problem that is usually encountered when testing MT4 expert advisors. What causes this error? It’s called ERR_INVALID_TRADE_VOLUME in the MT4 code. That means that your expert advisor is trying to send an order with invalid trade volume. On the absolute majority of the MT4 brokers setting some EA to open an order 0.123 lots will generate this error. But sometimes it’s generated when the EA, created for mini or micro accounts, is used on the standard account.

    I am using a micro live account, could it be that the EA is programmed with the wrong lot size for the account I have? Or does the .json exported from MT4 file know all that stuff automatically?

    Yes, I know my data is very limited.. it has been a huge struggle for me. My understanding is the the best data is the ACTUAL data from your broker, but here in the UK, not many brokers seem to offer MT4 that comes with data longer than a few months when you first install it.

    My method to get around this, so far, has been to set up a VPS purely for the purpose of collecting data – I first started collecting it in January 2018, but very frustratingly (and I hold my hand up here for negligence), the VPS must have been rebooted by the hosting provider, possibly for maintenance or similar, some time in June 2018, because after a very busy summer/autumn with my other business, I checked the VPS  in Jan 2019 to assess the data it had collected, only to find it had not been doing so since Jun 2018. Hence using the small amount of data I do have, starting in Jan 2019, and looking at the M1 timeframe, as that is obviously the highest count of bars I have for getting the largest data horizon.

    I read about people using services like Tickstory, but I just don’t see how it can be as reliable as ACTUAL broker data, do you have any thoughts on that?…

    In any case, after looking up the ordersend error 131, I altered the lot size for the strategy to 0.1 lot, and tried again in the MT4 Strategy Tester, and actually got much better results, please see below:

    (PS: this isn’t a strategy at all developed for trading, it’s purely for checking that my workflow process is being set up ok)

    You can see that it has 410 journal entries… the MT4 Strategy Tester results tab as 440 results, although the times and prices of the trades don’t match up very well at the start of the backtest, but they do at the end… why is that?

    Is it because EA Studio is not calibrated to the same timezone as my MT4?

    The graph is similar to the graph in EA Studio, although it shows better results than in EA Studio..

    I don’t know what much of the report tab means, but I’m guessing that the completely red bar at the top isn’t good?..

    Thanks,

    Simon

    #13767
    Admin bar avatarSimon
    Participant

    HI again Petko,

    With reference to the large spread on GBPNZD, I realise this, the spreads are much lower for other pairs with my broker, it’s just that GBPNZD moves a LOT of pips, and my thinking was that; if it can make really big moves, and EA Studio can make a strategy to work on it, even with the high spread, then it might be ok?..

    I will also look at other pairs in any case.

    Thanks,

    Simon

    #13802

    Hello Simon,

    Glad to hear from you. I have been over all the things you are struggling with right now, and I will assist you as much as I can because I know how hard it is.

    You are correct that the best data is the actual one from your broker. And yes, all the brokers provide small data. However, when you start collecting the data, you can download the hst files from time to time. This way you will make sure not to lose them again.

    We are about to provide Historical data from Ducas copy, we have done a lot of tests and we are at the final stage. This data seems to be very accurate and detailed. We tested creating EAs with it, and compared to most brokers, it is similar. Also, what we do is to convert it from tick data, which makes it very precise. Maybe it will be available next month.

    Stay away from the strategies that have no SL. They show great backtests but in actual they can drive you to huge losses.

    Now, EA Studio uses the timezone from the data that you have imported.

    Regarding the GBPNZD, you are correct but still, you don’t want to do algo trading with a huge spread.

     

     

     

    #13808
    Admin bar avatarSimon
    Participant

    Hello Petko,

    Thanks for your advice, I will look at strategies on other pairs and make sure there is a stop loss – what do you think about the lot size issue i mentioned? Does EA Studio know the difference between Micro, Mini and and Standard accounts? Increasing the lot size from 0.01 to 0.1 seemed to do the trick in terms of getting similar results in both programs?..

     

    I saw a video at the weekend where you go over how you set up your Reactor, I will follow those steps as a starting point, perhaps just using the 0.1 lot size.

    That is interesting about the Dukascopy data  you are developing, I will be pleased to look at it when it is available. In the meantime, I will begin saving my broker data on a regular basis. Do you have any tips for managing it? MT4 does start to cut off data after a while I think, so it would be more than a case of just replacing the previous file with a new one each time data is exported, so some work would be required in trimming off duplicate data and adding the new data to the old in chronological order, no? Or can it be managed with a spreadsheet somehow? Ideally, each timeframe for each pair would need to be kept in seamless chronological order in one file?

    Thanks again,

    Simon

    #13852

    Hello Simon,

    the EA Studio follows the Account settings you have in tools. And it follows the historical data that you use.

    It is good to keep it similar in EA Studio and MT Backtester to get the same results. Same starting account, same lot trading, same spread.

    Also, the MT4 backtest starts from 100 bar and there is such an option in tools, make sure to check it:

    Now regarding the data. MT4 should not cut the data, at least it does not with me. I am not sure if the broker can cut it..

    I just download these hst files from time to time to make sure it is not lost if I have to preinstall the MT.

    Also, in of my courses called Forex strategy course – Portfolio trading with 12 Expert Advisors

    I show how to save the bars in Excel which is the 100% option where you will save your data. But that works for FSB Pro. In EA Studio and json files are different and it is harder there to save them this way.

    Let me know if you have more questions.

    Cheers

    #13912
    Admin bar avatarSimon
    Participant

    Hello Petko,

    Thanks for your reply.

    I will begin saving my data using Excel asap. I suppose that even though we need it in .json format for EA Studio, we could import it into MT4 should it be necessary, and run the EA Studio script as normal.

    When I have things organised, I will certainly look at buying a course. Thank you for all your help so far.

    Kind Regards,

    Simon

    #13999

    Hello Simona,

    actually, the idea of the courses is to help you organize your work with the strategy builders. This is what I teach – how people to get started an not to lose time struggling around.

    Anyway, do not hesitate to ask me things.

    Cheers,

    #14004
    Admin bar avatarDesita
    Participant

    Simon, I have all courses from Petko, and I enroll in every new one that is being launched. You can not imagine how much they helped me to become a profitable trader (from a loser).

     

    #14025
    Admin bar avatarSimon
    Participant

    Hello Petko and Desita,

    I will be happy to invest in the courses.

    My trouble is that I only have data from 10th of January 2019. Until that problem is solved, I don’t think I can get anywhere.

    What Petko mentioned above about possibly providing Ducascopy data in the future could be the answer, but until then, or I have had enough time to successfully accumulate seamless data to work with, the best I can do is familiarise with the software, I think?

    In the meantime I have other things to consider, such as how I will go about running the generator for ten hours a day?, should I buy a dedicated laptop for that purpose or do I use a VPS for it?, how much extra will my electricity bill be if I use a laptop?, should I still use my VPS for actual trading with the strategies?, etc, etc.

    I am sorry for all the questions, I have so many!

    Thanks,

    Simon

    #14176
    Admin bar avatarDesita
    Participant

    Hey Simon, most of us had these same questions, and Petko is always giving the answer, so now I will save him some work 🙂

    Regarding the data, why don’t you look for another boker with more bars? I remember Jacpin shared a broker that provided 5 years of data, and she was happy with the data. I forgot the name. The other option is to use the MetaTrader-Demo data that is by default in MT. This data is very accurate. I see that Petko uses it in many videos.

    Normally I use my laptop and desktop to do the generation and the actual trading. I do not use VPS. A VPS is an extra cost that you need to pay – about $15 per month minimum. I do not think that one computer or a laptop will be such an expense in electricity.

    If you have a laptop just leave it running overnight. There is no problem with that.

    Regards,

     

     

Viewing 15 posts - 1 through 15 (of 15 total)
  • You must be logged in to reply to this topic.