How to backtest a gold EA in MT5: a practical test plan

Create a reproducible MT5 test with documented costs, modelling assumptions and a separate period for checking selected settings.

All guides

At a glance

Can this EA build run in Strategy Tester?

Confirm tester support with the product's current instructions before starting. MQL5's WebRequest function cannot execute in Strategy Tester. An EA that normally contacts a licence or data service may therefore need a supported testing workflow. Gold EA's usual connection instructions are not proof that every supplied build can be backtested.

Ask support which approved build and setup to use if this is unclear. Do not invent a licence bypass, edit the executable or assume a run with no trades is a valid strategy result. Save any initialization error as a compatibility question to resolve first.

How do you make a test reproducible?

Prepare a run sheet before comparing results. Record the EA version, MT5 build, broker server, full symbol, timeframe, dates, deposit currency, initial deposit and input file. State what question the test is intended to answer, such as whether a documented setting change behaves consistently across periods.

Keep one baseline and change one assumption at a time. Use names that identify the period and configuration, then save the report and logs together. If another person cannot reconstruct the run from this record, the headline result is less useful than it appears.

Which historical modelling mode should you inspect?

MT5 offers generated-tick, real-tick and simplified price modes. Record the selected mode and inspect the available history rather than treating every mode as equivalent. Real-tick mode uses broker tick history; that does not reproduce every condition of a future live account.

Simplified modes can change execution behaviour compared with more detailed modes. Review the tester's modelling limitations when intrabar behaviour matters. Missing data, corrections or a large result difference between modes deserve investigation. Do not select a mode solely because it produces the strongest curve.

How should you account for costs and delay?

Check commission settings and the configured execution delay in MT5. The tester supports predefined or custom commissions and can simulate delays for EA trading requests. Spread uses historical data under the tester's documented rules. Keep the precise assumptions attached to the result.

Compare the baseline with clearly labelled less favourable cost and delay assumptions where the supported workflow permits them. Include overnight costs and any separate operating charges in your review. This is sensitivity analysis: it asks how dependent a result is on assumptions, without claiming that one stress scenario covers every possible execution problem.

How do you avoid testing only what you fitted?

Choose a separate checking period before using results to select inputs. MT5's forward-testing tools separate a later portion of history for checking optimization candidates. That historical forward period is different from running the EA prospectively on a demo account.

Retain the selected inputs when you inspect the reserved period. If you repeatedly alter them after viewing that period, you are using it for selection too. Keep a record of rejected configurations and the number of trials. Otherwise the final report conceals how many attempts preceded the attractive result.

What should happen after the test?

Review the balance and equity curves, loss sequences, open exposure and logged errors with the settings frozen. Summarize the differences across periods and cost scenarios. A result that cannot be reproduced or depends on unresolved data errors is not ready to guide the next operational step.

Use a compatible demo setup to observe current behaviour separately from historical simulation. Confirm account assignments before Gold EA activation because paid account slots are permanent. A backtest is a documented experiment, not a promise of live execution or profitability; no testing process eliminates the possibility of losing money.

Sources & further reading

Official documentation helps you confirm how the terminal works. Product-specific settings should follow your current Gold EA guide.

MQL5 Reference: WebRequestMetaTrader 5 Help: Strategy TestingMetaTrader 5 Help: Testing FeaturesMetaTrader 5 Help: Strategy Optimization and Forward Testing

Product education from Gold EA Editorial. Trading involves risk; this guide does not predict returns or provide personal investment advice.

Editorial standards and corrections →

Related guides

Build on what you’ve learned.

See how Gold EA fits your setup.

Explore the EAs, account dashboard and setup guides before choosing your access.

Explore Gold EA