EA setup in MT5: practical checks before trading

Start with the complete preset

Save the full input set before importing a partial risk file. Check the actual broker symbol, including suffix, contract size and digits. The chart timeframe may differ from the signal timeframe inside the EA. Treat another symbol or timeframe as a separate setup requiring evidence rather than assuming the old result transfers unchanged.

Verify protective behavior

Check initial stops, volume limits, free margin and rejected-request handling. A terminal receiving a trade request does not prove execution. Keep return codes and confirm actual positions. Test what happens when data or an indicator is unavailable; silently switching off a trading filter changes the strategy and makes the previous test an unsuitable reference.

Check the account as a system

Use distinct magic numbers where required and verify position attribution. Netting and hedging can behave differently when several EAs use one instrument. Restart with open positions and pending orders to inspect recovery and duplicate-entry prevention. Use isolated terminals for tests; do not stop an active trading terminal merely to free a test instance.

Checks to make before deciding

• Describe the broad entry idea, exit mechanism and intended market conditions. Keep an internal precise specification even if the public explanation omits implementation details.
• Check indicator shifts, buffer readiness and bar closing times. Confirm that an entry uses information available at that moment, not the completed value of a future bar.
• Test repainting and revised external data. Preserve what was known at signal time and compare it with later values; a clean historical plot alone is insufficient evidence.
• Inspect whether positions are added by distance, loss recovery or exposure scaling. Support no-grid and no-martingale claims with all relevant code paths, including shared components.
• Compile the final sale source without errors and review warnings. Archive the exact compiler output and compare binary checksums with the version used in final testing.
• Test netting and hedging only where claimed as supported. Examine opposite orders, partial fills, reversals and magic-number attribution under each relevant account model.
• Restart with open positions and pending orders. Verify state recovery, no duplicate entry, correctly restored stops and safe handling of missing history or indicators.
• Simulate connection loss, market closure and rejected trade requests. Check return codes and bounded retries; do not treat submission as confirmed execution.