Strategy Tester fails with custom .set file after build 6180 update

New comment
 

Hello! I'm having a reproducible problem with the Strategy Tester on MetaTrader 5, after the automatic update to build 6180 (was build 6140 before).

The problem: Any backtest (single pass or optimization) that uses a custom Expert Parameters (.set) file fails instantly, with the result "some error after pass finished", taking 0:00:00.000 — meaning the test never actually runs, no trades are simulated at all.

Most importantly: this happens even with a .set file that was working perfectly BEFORE the update — an EA that has been running live for weeks, using the exact same file and parameters. It only broke after the MT5 update to build 6180.

When I run a backtest WITHOUT any custom .set file (just the EA's default input values), it works normally. The problem is specifically when loading a custom .set file via ExpertParameters.

What I've already tried, without success:

  • Fully reinstalling MetaTrader 5 (official mt5setup.exe installer, verified signature)
  • Recompiling my EA under the current build (0 errors, 0 warnings)
  • Closing all open charts in the terminal
  • Testing with different symbols and deposit amounts
  • Testing on a clean account state, no open positions
  • Disconnecting/reconnecting the account (happens automatically on every terminal restart)
  • Testing via config .ini file (headless) and manually clicking "Start" in the Strategy Tester (both fail identically)

Technical details:

  • Server: RoboForex-ECN
  • MetaTrader build: 6180 (was 6140 before the auto-update)
  • Windows 11 build 26200
  • RoboForex support confirmed they don't control the Tester/terminal and recommended reporting here

Has anyone else experienced this after updating to build 6180? Is there a known fix, or a way to request the previous build (6140) while this gets resolved?

Thank you!

 
Subject: MT5 5.00 Build 6180 Command-Line Backtesting Error: "The procedure entry point WorldTransform could not be located in the dynamic link library"

Issue Summary:
When attempting to backtest via command line,  terminal64.exe  crashes immediately upon startup with exit code 0xC0000139 ( STATUS_ENTRYPOINT_NOT_FOUND  - The procedure entry point could not be located in the dynamic link library).
Specific Symptoms:
  • Launching  C:\Program Files\MetaTrader 5\terminal64.exe  via command line results in a crash within 0.1 to 37 seconds.
  • Exit code: 3221225785 (Decimal) = 0xC0000139 (Hex).
  • No logs are generated (no new files in the  logs  directory), indicating the crash occurs during early process initialization, before the main application logic loads.
  • Unable to run the Strategy Tester.
Troubleshooting Steps Attempted (In Order):
  1. Version Switching: Tested builds 6180, 5833, and 5200. All crashed identically. MT5 forces an auto-update back to build 6180.
  2. Portable Mode: Tested with and without the  /portable  flag. Both result in crashes.
  3. Config Encoding: Changed  .ini / .set  file encoding to UTF-16 (matching previous successful configurations). No effect.
  4. Registry Cleanup: Removed  WIN8RTM  compatibility flags from  AppCompatFlags\Layers . No effect.
  5. Account Data Check: Verified account data exists (confirmed). Not an account issue.
  6. DLL Hijacking Check: No private DLLs found in the MT5 directory;  PATH  environment variable is clean. Ruled out.
  7. Global Injection Check: Checked  AppInit_DLLs  and IFEO (Image File Execution Options). Both are empty. Ruled out.
  8. Static PE Analysis: Analyzed the import table of  terminal64.exe . All statically imported functions from  GDI32.dll ,  gdiplus.dll , etc., exist on the current system. No missing dependencies found.
  9. Crash Dump Configuration: Configured  LocalDumps  to capture crash dumps. The crash occurs too early; WER (Windows Error Reporting) fails to generate even a minidump.
  10. Digital Signature Verification: Official MetaQuotes signature is valid.
  11. Launch Method Comparison: Found a critical discrepancy:
    • Launching via  subprocess  (Python/bash) -> Always crashes.
    • Launching via  Start-Process  (PowerShell) -> Survived once, but subsequent identical tests crashed again (intermittent).
Core Contradiction (Critical for Official Support):
On the same machine, with the same executable, and using the same launch method, the results are unstable:
  • One instance of  Start-Process  (no arguments) allowed the process to survive for 5 seconds.
  • Immediately repeating the exact same operation resulted in consistent crashes with error 0xC0000139.

💡 Additional Note for Your Report

Based on your previous screenshot showing the garbled character "酃WorldTransform", you might want to add this specific observation to the report, as it strongly suggests a binary corruption or encoding mismatch in the specific build installed on your machine, rather than a missing system DLL.
Suggested addition to the report:
Observation: The error dialog displays a garbled character prefix (e.g., "酃WorldTransform") before the function name. This suggests potential binary corruption in the installed  terminal64.exe  or a locale/encoding mismatch during the auto-update process, rather than a genuine missing system API.
 
Osmar pinheiro:
Has anyone else experienced this after updating to build 6180?
I once experienced the same issue in a previous Build. I was able to "force" the settings into the Tester anew by resetting and resaving the settings in the Tester window─not in the Expert window. It appears from your OP that you did most everything except recreate the set file.
New comment