- This topic has 16 replies, 2 voices, and was last updated 1 year ago by
.
-
Posts
-
Hello,
somewhere in the learning videos I heard that I should test different brokers with same EAs.
The problem is, when I create an EA with Expert Advisor Studio, it is customized for a demo account of a broker.
If I were to test the EA in a live account or with a different broker, then the EA would probably stop working. After all, each broker has its own conditions (spread, timezone, commissions, etc.).
So how do I test the same EAs at different brokers?
Greetings
Achmidius
Hi Achmidius, welcome to the forum:-)
Okay, so first of all ideally your EA should be robust enough to work well on most brokers and demo or live. An EA that only works on one set of broker data would not be good strategy to trade.
This is very straight forward to quickly test in EA studio thankfully.
Just open up a demo account with any other brokers you want to test your EA on. The upload their data to EA studio and then backtest your EA using the different broker data.
Your Ea strategy should perform in a similar fashion generally on the different brokers data. If it works well on one but poorly on another then that’s a red flag for me.
Also, you can do this test importing the data to compare between the live and demo terminals for your broker. What broker are you using?
As for the spread commissions and swaps. Use 10 spread for raw accounts and 20 spread for non commission accounts.
The current swap values will import automatically to EA studio but I would suggest selecting a strategy and then test changing the values on the symbol settings to see how the strategy statistics change, likely you will find that your profit will not be very sensitive to the swap values for most strategy’s.
Timezones are automatically accounted for when importing data from Mt4 to EA studio of course so in doing these checks using different broker data that is all taken care of.
Thank you for the very useful tips. These will help me a lot.
>What broker are you using?
I currently use Dukascopy demo account.
The videos made me realize how important it is to test multiple brokers.
What I do not realize is whether later after my selection of a broker, the EAs are then always more profitable than with other brokers.
At the moment I do not follow any strategy. All that matters to me is that there are no swap fees.
My set times: Monday to Friday from 01:00 – 23:00 o’clock
At the same time EA Studio runs very slowly for me. 13 EAs after 2 days.
Is there any way to speed it up through configuration?Greetings
Achmidius
Happy to help Achmidius,
There is a good list of recommended brokers on the blog btw so I would recommend checking them out.
I like IC markets as easy to open lots of demo accounts and they have tight spreads in raw account but there’s plenty good brokers out there.
Dukascopy is good but their demo accounts expire after 2 weeks which can be limiting, I have a non expiring one that I got when I opened a live account but usually they expire.
I would definitely recommend trying a couple of other brokers out for demo trading. If you follow what I suggested in previous post then as long as you have checked that the strategy is working similarly on that broker over a sensible period then all good to trade on it (demo first of course, don’t rush into live trading until have several positive months of demo trading under your belt)
Can I ask why are the swap fees so important to you Achimidius?
Also regarding your small amount of EAs in the collection. What settings are you using to generate? It sounds like your acceptance criteria could be too strict?
Specifically can you share timeframe, acceptance criteria and any robustness checks you are performing?
Also what is the number of generated strategies rather than those making it to the collection? That number is more of an indication of how fast EA studio is working, the small number added to the collection is more to do with you overly strict settings.
I can’t write anything more specific because the forum has protection from Wordfence or something like that.
Block Reason:
A potentially unsafe operation has been detected in your request to this site.Hmm, that sounds odd.
Are you able to complete the following for example (if not will need to talk to Petko about what could be causing the problem):
1. Symbol – EURUSD H1 50000 bars
2. Robustness – Monte Carlo only 80/20
3. Acceptance – complete: pf> 1.2, is: pf> 1.1, oos>1.1
Can you replace the info above with your values? Also how many calculations are being performed (the number right of the start button)?
I am almost positive that the issue of so few strategies in the collection is either an overly strict acceptance criteria, robustness check or a combination of them both.
The information on the reactor screen will show what the bottleneck is ie if you see after a few hours that you have 1000 EAs passed to Monte Carlo check but only 2 make it through that check to the collection then that is a problem with Monte Carlo.
Or perhaps you see that of the 200,000 calculations on 2 are making it through the acceptance criteria and then being passed to Monte Carlo, the you know your acceptance criteria is too strict.
Hope that makes sense and you get to the bottom of the issue soon? If not then let me know and there may be a couple of other things but this would be where I would start troubleshooting
I have a similar setup with M15, but without Monte Carlo. I tried to transfer the settings from MT4 to “Symbol Settings”. So very few EAs are created for me.
Unfortunately I don’t know what is the trigger for the block of Wordfence. I sent an email to info@eatradingacademy.com yesterday with the text I would post here in the forum regarding EA Studio settings and the reason for swaps. So far I have not received a reply.
Okay, so to my understanding you are using about 50000 bars on M15 with and no reactor.
It sounds like an overly strict acceptance criteria. I see you have set some time settings and say Mon – Fri, does this means you have the check box ticked to close trades on Friday.
I would suggest taking a few screenshots of your acceptance criteria, time setting and the generator after running (to see number of generations and added to collection) and posting here.
But my immediate guess is that if you have restricted the time settings that you probably have too many minimum trades set in the acceptance criteria.
How many minimum trades do you have set? You could initially try removing minimum trades from acceptance criteria and then use the performance filter in the collection to see what is a more reasonable value to use given your time settings.
Of course without more information I am guessing and speculating. Screenshots and more precise information are really needed. I am sure the wordfence block issue will be sorted soon so we can properly resolve.
Another suggestion around this issue is I suspect it is a local issue to you, you mentioned you were using the brave browser have you tried logging into the forum using Chrome instead??|
>Can I ask why are the swap fees so important to you Achimidius?
Because of private and commercial trade.
I once read years ago that a private trader in Germany had to go to court because of swap fees. Whether he was acquitted then, I don’t know anymore.
I just do not want to give the tax office reasons to classify me as a commercial trader 🙂>Specifically can you share timeframe, acceptance criteria and any robustness checks you are performing?
Settings for Dukascopy Demo account:
>Also how many calculations are being performed (the number right of the start button)?
I don’t know anymore. Windows updated my computer overnight.
Best regards
Achmidius
Okay perfect, now we are getting somewhere.
So basically you are working with 4 years of data on EURUSD M15. First point is that I would get in the habit of checking the box “use end date” on the data horizon tab.
This is as I expected, the combination of your time settings and all the boxes closing the sessions checked with such a high minimum trade count is definitely pretty strict.
So a few points:
Obviously 4 years of data with M15 requires more bars than say 4 years of H1 so that you can definitely expect less calculated strategies in the former case.
But I think that with your time restricted setting and such a high minimum trade count over 4 years that this is resulting in a pretty fair small amount of strategies passing this acceptance criteria.
If this is what you are after then fair enough but I would suggest reducing the minimum trade count to 250 for example and then once more strategies are in the collection you can use the performance filter.
I’m sure you will see in the generator that plenty strategies are being calculated but very few are being ascended. Essentially for your generator settings the acceptance criteria is too strict.
Also remove all the “minimum net profits =10 “ in the acceptance criteria.
This is redundant since you have set the profit factor about 1 in compete, is and oos.
I would say it is good practice generally if struggling to get enough strategies to make acceptance criteria less strict and then use performance filter.
Petko has said it before, much better to have lots of strategies in the collection that you can filter than only a handful of strategies right?
Hello,
with the new settings, more EAs are generated now.
I have opened 5 demo accounts on a VPS server and started 22 EAs per demo account.
I think the principle of EAs is not clear to me yet.
Can several EAs open a position on a demo account at the same time? If so, then the capital would be wiped out in no time, wouldn’t it?I would appreciate an explanation.
Best regards
Achmidius
You must be logged in to reply to this topic. Login here