New MetaTrader 5 build 2360: Extension of SQLite integration

To add comments, please log in or register
MetaQuotes
Moderator
216285
MetaQuotes  

The MetaTrader 5 platform update will be released on Friday, March the 6th, 2020. The new version features the following updates:

  1. MQL5: SQLite database operation errors can now be analyzed using standard MQL5 tools. The following error codes have been added:

    • ERR_DATABASE_ERROR — generic error.
    • ERR_DATABASE_INTERNAL — SQLite internal logic error.
    • ERR_DATABASE_PERM — access denied.
    • ERR_DATABASE_BUSY — database file locked.
    • ERR_DATABASE_LOCKED — database table locked.
    • ERR_DATABASE_NOMEM — insufficient memory for completing operation.
    • ERR_DATABASE_READONLY — attempt to write to read-only database.
    • ERR_DATABASE_IOERR — disk I/O error.
    • ERR_DATABASE_CORRUPT — database disk image corrupted.
    • ERR_DATABASE_FULL — insertion failed because database is full.
    • ERR_DATABASE_CANTOPEN — unable to open the database file.
    • ERR_DATABASE_PROTOCOL — database lock protocol error.
    • ERR_DATABASE_SCHEMA — internal use only.
    • ERR_DATABASE_TOOBIG — string or BLOB exceeds size limit.
    • ERR_DATABASE_CONSTRAINT — abort due to constraint violation.
    • ERR_DATABASE_MISMATCH — data type mismatch.
    • ERR_DATABASE_MISUSE — library used incorrectly.
    • ERR_DATABASE_AUTH — authorization denied.
    • ERR_DATABASE_RANGE — bind parameter error, incorrect index.
    • ERR_DATABASE_NOTADB — file opened that is not database file.

  2. MQL5: Fixed operation of the DatabaseImport function which enables data import from a file into a database table.
  3. MetaEditor: Fixed output of strings longer than 32KB to logs.
  4. MetaEditor: Fixed occasional incorrect encoding in messages sent from the Python console (stdout, stderr) into the Errors section of the editor.
  5. Updated documentation.

The new MetaTrader 5 version will be available through the LiveUpdate system.

Daniel Weckhuyzen
196
Daniel Weckhuyzen  

Hi,


Thx for the new release.

I've seen a bug that also appeared before in earlier releases in the "Market scanner" part of the backtester.

After a scan I see again the same symbol appearing more than once.


Regards,

Danny

Files:
doubles.png 44 kb
Sergey Golubev
Moderator
115560
Sergey Golubev  

Forum on trading, automated trading systems and testing trading strategies

Learn Russian

Sergey Golubev, 2020.03.08 07:11

It is difficult to learn.
You can read/write on the Russian forum using the automatic translation tool which exists on every post (many users from English/Chinese/etc parts of the forum are using this automatic translation tool).

If you really want for some post (technical/programming issue/question) to be sent to the Russian forum (to the threads where the admins are replying on technical questions) - so ask the users to pass your post to Russian forum (because some moderators and users can read Russian/English, and many members are using automatic translation tool with no problem at all).

But it is much more better if you will use this automatic translation tool by yourself without asking anyone (on the same way as many people are using it).


Daniel Weckhuyzen
196
Daniel Weckhuyzen  

Hi,

I've discouvered a new bug in the backtester. I've added a url with a video that demonstrates the problem.

There's a difference in results when backtesting over a 1 day or 2 days period.

To resume what happens in the video:

first run: using OHLC over 1 day: the backtest shows 2 trades

second run: using OHLC over 2 days: the backtest shows for the most recent day 6 trades

third run: using OHLC (same as first run) : the backtest shows 6 trades

fourth run: using real ticks over 1 day: the backtest shows 2 trades

fifth run: using real ticks over 2 days: the backtest shows for the most recent day 6 trades

sixth run: using real ticks over 1 day (same as fourth run): the backtest shows 6 trades.

I hope the video will clarify the problem.

I've also seen a huge difference between OHLC backtesting and real-tick backtesting; it's like the indicators are wrongly calculated


PS: I've seen the same problem also in the latest beta release 2363




http://users.telenet.be/karaoke/backtest_problem.mp4

Daniel Weckhuyzen
196
Daniel Weckhuyzen  

Hi,

I've noticed another problem in the backtester and optimiser.

I've been setting breakpoints during visual backtesting and I 've seen that the "commission" costs are still not calculated in.

It's not possible to have reliable backtests when this does not work as expected.

I have tested 'real ticks' and OHLC; both have the same problem.

I have also added a url that demonstrates the problems

http://users.telenet.be/karaoke/commission_problem.mp4



Thx for your quick reply !


Danny

Testing trading strategies on real ticks
Testing trading strategies on real ticks
  • www.mql5.com
The article provides the results of testing a simple trading strategy in three modes: "1 minute OHLC" using only Open, High, Low and Close prices of minute bars; detailed modeling in "Every tick" mode, as well as the most accurate "Every tick based on real ticks" mode applying actual historical data. Comparing the results allows us to assess...
Files:
Marcel Fitzner
393
Marcel Fitzner  
Any update on this? Or is it still an open issue?
To add comments, please log in or register