BUG on optimization "All symbol in MarketWatch" option

 

Hi everyone, I have always noticed a problem related to the optimization of the strategy on all symbols. Almost always the results obtained in the report of a symbol were different than the single execution on that symbol. I never explained why and it forced me not to be able to take advantage of this feature and to perform the executions individually on each currency.

I think I understand why: for some reason, the time interval of the data is altered and the execution of the strategy during optimization occurs over a different period (usually shorter) than the one selected.

In fact, I needed data for some processing on multiple currencies, so I created a script for writing the daily data to file: if I ran the script individually, the time interval was correct, if instead I ran the script on the entire market whatch, the files contained only the last two years.

Please investigate on that.

 

Investigate on what ? You don't provide any technical details, no code, no screenshots, no log, nothing.

From my experience there is no problem with "All symbols of the Market watch".

 

Hello Alain,

attached the the code I used to export the data.

If I ran the code on a single symbol the result was fine. If I ran on all the market watch, data in the files start from a nearest time.

For example, running on "Coca-Cola" the file is correct and data starts from 2015. (see the attached "EXPORT_CANDLE_Coca-Cola.CSV.good" file)

Running the script on all market watch, the file produced for "Coca-Cola" starts from 2020/08 (see the attached "EXPORT_CANDLE_Coca-Cola.CSV.bad" file)

Unfortunately this behaviour is not deterministic, I mean not always coca-cola produce a bad file after multiple execution. I suggest to discover the bad files looking the file size produced: the bad files have a small size.

This is a post a wrote in the past for a similar problem: https://www.mql5.com/en/forum/348522

I think the problem is not only on 'all market watch' but also during normal optimization.

The broker I use is trading.com.

Let me know if you need other info.

Bug in "All Symbols selected in MarketWatch" optimization
Bug in "All Symbols selected in MarketWatch" optimization
  • 2020.08.07
  • www.mql5.com
Hi All, using the optimization "All Symbols selected in MarketWatch" I noticed some different values between the Optimization Result related to one...
 
Francesco Pinta #:

Hello Alain,

attached the the code I used to export the data.

If I ran the code on a single symbol the result was fine. If I ran on all the market watch, data in the files start from a nearest time.

For example, running on "Coca-Cola" the file is correct and data starts from 2015. (see the attached "EXPORT_CANDLE_Coca-Cola.CSV.good" file)

Running the script on all market watch, the file produced for "Coca-Cola" starts from 2020/08 (see the attached "EXPORT_CANDLE_Coca-Cola.CSV.bad" file)

Unfortunately this behaviour is not deterministic, I mean not always coca-cola produce a bad file after multiple execution. I suggest to discover the bad files looking the file size produced: the bad files have a small size.

This is a post a wrote in the past for a similar problem: https://www.mql5.com/en/forum/348522

I think the problem is not only on 'all market watch' but also during normal optimization.

The broker I use is trading.com.

Let me know if you need other info.

I tried on some broker with Forex symbols and I get the same data files (trading.com doesn't allow to easily open a demo account).

So it depends of the symbols in your Market Watch. Please provide the log file of your "All Market Watch symbols" optimization.

 

Yes, on forex symbols it was ok also for me. The problem seems appears on symbols with low resolution market data: the stocks symbols have only daily data before the 2020.

Attached there is the log file for the optimization a made.

About the broker, you can use XM Global since I think tranding.com is a european brach and symbols and data on XM are the same of trading.com.

I've tested also XM and the issue still there.

Files:
20211224.log  409 kb
 
Hi guys, has anyone got to check?
Reason: