EA works great on a Strategy Tester but it is not on a real account

New comment
 
I have done a great EA which gives me excelent results when simulating on strategy tester of mt4. On real account it doesn't give me the same results...
Files:
simula5fo_2017.gif  9 kb
 
NewbieForex: EA which gives me excelent results when simulating on strategy tester of mkt4. On real account it doesn't give me the same results...
  1. You over optimized the parameters, so it gives excellent results on ST. Market does not repeat exactly.
  2. variable slippage
  3. variable spread
  4. stops close at market (#1 and #2) Tester closes exactly at the stop.
  5. Price gaps inter-bar.
  6. It can take seconds to open an order (#1 and #2) During news, it can takes minutes. Tester is instantaneous. Connection problems, more minutes.
 

<Deleted>

 
Please write in English on this forum. Thanks.
 
whroeder1:
  1. You over optimized the parameters, so it gives excellent results on ST. Market does not repeat exactly.
  2. variable slippage
  3. variable spread
  4. stops close at market (#1 and #2) Tester closes exactly at the stop.
  5. Price gaps inter-bar.
  6. It can take seconds to open an order (#1 and #2) During news, it can takes minutes. Tester is instantaneous. Connection problems, more minutes.

Thank you

 
Helder Castro:

Thank you

Hello Whroeder

Thanks for your comments. Answering to your comments:

1. The EA is not optimized at all; 

2. I've done some variations on the slippage and didn't get any difference on the Strategy Tester; I don't know if it really affects or not;

3. Variable spread can be a problem, yes. The problem is that when changing the account type to one with constant spread, the spread is bigger. My account has a variable spread between 9 and 11 points. When simulating with a greater spread, it gives profits until a spread of 14. From 15 the ST only gives losses. So I don't understand if on a real account the variable spread has effect or not.

4. I've a stop of 30 points. I think its normal...

5. ? Please can you explain what you mean?

6. That's what I think its the main reason. How can I eliminate this? Do you think having an account with a broker nearest the stock market in Frankfurt will solve the problem? But I've search the net and didn't find any with the platform Mt4. Appreciate if you know any good Broker.

It was very nice to have your comments. Thank you.

 
Helder Castro:

Hello Whroeder

Thanks for your comments. Answering to your comments:

1. The EA is not optimized at all; 

2. I've done some variations on the slippage and didn't get any difference on the Strategy Tester; I don't know if it really affects or not;

3. Variable spread can be a problem, yes. The problem is that when changing the account type to one with constant spread, the spread is bigger. My account has a variable spread between 9 and 11 points. When simulating with a greater spread, it gives profits until a spread of 14. From 15 the ST only gives losses. So I don't understand if on a real account the variable spread has effect or not.

4. I've a stop of 30 points. I think its normal...

5. ? Please can you explain what you mean?

6. That's what I think its the main reason. How can I eliminate this? Do you think having an account with a broker nearest the stock market in Frankfurt will solve the problem? But I've search the net and didn't find any with the platform Mt4. Appreciate if you know any good Broker.

It was very nice to have your comments. Thank you.

I don't think @whroeder1 is present on the forum anymore, so I will try to answer your queries in his place.

  1. OK!
  2. Unless you are using a 3rd arty tool like Tick Data Suite, you will not be able to simulate slippage in the Strategy Tester. The closest you can come to this, is by using Virtual Stops (not real S/L or T/P) and only Market Orders (no Pending Orders) in your EA. You can also change your code to purposely delay executions so as to simulate slippage.
  3. Also, unless you are using a 3rd arty tool like Tick Data Suite, you will also not able to simulate variable spreads in the Strategy Tester. Even if you use a higher spread in testing, it will still give incorrect results because it remains constant. The variability of the spreads between the open and close of the positions is what can cause greater losses if the strategy is not properly defined or if the EA is not properly coded.
  4. This point has nothing to do with your stop-loss size or the Stops Level defined by your broker, but has to do with what I have already mentioned above, that you should use virtual stops and no pending orders in order to get more realistic results. This is because the strategy testers causes the Stops and Pending orders to be triggered at the exact price they are set, instead of of respecting the market price at the time.
  5. The pseudo tick data generated by the Strategy Tester is not very realistic. Real prices can have huge gaps (sudden jumps in price). You can only experience these realistic events in the Strategy Tester when you use real tick data when using 3rd party tools like TickStory or Tick Data Suite.
  6. Even if you rent a VPS from MQL5 or via 3rd party ISPs, in order to have a faster, closer connection to the broker, you will still experience some communication latency in general as well as large delays during news events. None of which are simulated in the Strategy Tester for MetaTrader 4. You can however, make changes in your code in order to simulate such latency and delays during news events. Also, you don't have to search for VPS with MetaTrader. You can simply use normal VPS providers for a Windows VPS and then install and maintain MetaTrader 4 yourself. Since we are not allowed to discuss commercial sites on the forum, you can contact me via a PM and I will give you a VPS provider suggestion.
 
Fernando Carreiro:

I don't think @whroeder1 is present on the forum anymore

what happen to him?

 
Mohamad Zulhairi Baba: what happen to him?

I'm not really sure. I did notice that he had been temporarily banned for a short time, but I don't know the reason. He probably decided that it was not worth coming back after that!

New comment