Open View → Strategy Tester or press Ctrl+R. In Settings, select the compiled Expert Advisor already available in your platform.

Select the full Symbol and Period (chart timeframe). Enable Use date and enter From and To to restrict the historical range.

Choose Model: every tick, control points or open prices only. Open-price testing does not reproduce all movements within a candle; record the selected method with the result.

Check Spread. The default uses the symbol’s spread at the start of testing to model Ask prices from Bid history; a custom spread is also possible. This is not continuously changing historical spread.

In Expert properties → Testing, set the simulated initial balance, currency and allowed trade directions. Under Inputs, set each Value; the Start, Step and Stop ranges apply to optimisation.

Leave Optimization off for one fixed-parameter test. Optionally enable Visual mode, then click Start. Afterwards inspect Results, Graph, Report and Journal, including modelling quality and errors.

In Tester → Settings, select an available EA, symbol, timeframe, modelling method, date range and spread. Enable Optimization; leaving it disabled runs a single fixed-input test.

Open Expert properties → Testing. Set the simulated deposit, currency, permitted trade directions and ranking criterion. Choose genetic search only if intended.

In Inputs, check each parameter to vary and set its Start, Step and Stop fields. An unchecked parameter retains its Value.

In Expert properties → Optimization, enable and configure only the pass limits you intend. Reaching a limit interrupts that pass; the next pass continues.

Select Start. Inspect Optimization Results and Optimization Graph; results summarise passes, not every trade. Sort by a column and check whether Skip Useless Results is hiding unprofitable passes.

Double-click a pass or choose Set Input Parameters. MT4 restores its inputs to Settings and disables Optimization. Check the restored values, then select Start for a single test.

In Navigator, open the EA’s context menu and choose Test. In the tester’s Settings, choose a single test rather than parameter optimisation.

Choose the main Symbol, timeframe and Date range; the end date is excluded. For a multi-symbol EA, enable its required instruments in Market Watch → Symbols.

Review the tick-modelling mode and execution delay. Generated ticks and available real ticks are different data choices; a no-delay test represents ideal execution rather than evidence of live performance.

Set the simulated initial balance, currency and leverage, then enter the EA’s Inputs. Record these assumptions alongside the test; they describe the simulation.

For chart playback, enable visual testing with a local agent and optimisation off. Click Start in Settings; history downloads may take time.

Review Result, Graph and Journal. Check data-related messages and any shifted start date, then keep the report with the EA version and settings. Read the interpretation below before comparing results.

Open the EA’s Test command in Navigator. In Settings, confirm the symbol, timeframe, dates and simulation assumptions.

Select parameter optimisation rather than a single test.

In Inputs, tick parameters to vary. Set Start, Step and Stop; unticked parameters remain fixed. Review the displayed combination count.

Choose a search mode and, for genetic optimisation, a ranking criterion.

Set Forward before starting if required, using the instructions below.

Click Start. After completion, sort Optimization Results by a column heading. Double-click a pass to test its parameters in a single run.

In Settings → Forward, select 1/2, 1/3, 1/4 or Custom.

The later part of Date is reserved; optimisation uses the earlier part. For Custom, set its start date.

After completion, compare selected passes in Optimization Results and Forward Results. Not every combination is necessarily forwarded.

Keep both periods and parameter sets with the reports. Re-tuning on the forward period removes its independence.

Allow the required symbol history to synchronise. A multi-symbol strategy may pause while other history is loaded.

Compare requested and actual dates in Journal. Insufficient pre-start data can shift the start forward.

Retain that shift in the report; do not present the requested period as fully tested.

Changing timeframe to shorten preparation changes the experiment. Check the EA’s design before comparing results from that different timeframe.

After starting the visual test as described above, use the separate visualisation window. Its main chart uses the testing timeframe; choose that timeframe before starting because it cannot be changed inside this window.

Read the chart's trade markers alongside its test prices. These show simulated activity, not trades or prices currently occurring in your Live account.

Open View → Market Watch or press Ctrl+M to inspect the prices and current time within the simulation.

Open View → Data Window or press Ctrl+D. Move the pointer over a bar to inspect its date, OHLC prices, spread, volume and indicator values.

Open View → Toolbox or press Ctrl+T. Compare Trade for simulated positions and pending orders, History for orders and deals, Operations for EA requests, and Journal for testing logs.

When investigating an unexpected action, match the simulated timestamp across the relevant request, execution and log entries. A requested operation and an executed deal are different records.

Read Journal chronologically. History preparation may download the main symbol and currency-conversion symbols; data preparation is different from an EA already running.

An OnInit failure means the EA could not initialise. Check its inputs and the accompanying message; this is not evidence of a rejected live order.

Memory errors, an array index outside its range and division by zero can terminate a test. Record the exact error and discuss it with the program developer rather than treating a partial report as a completed test.

Compare the requested date range with the actual stopping point. Simulated Stop Out or the EA removing itself can end the run early. Keep the EA version, settings and relevant log messages with the report; an early end alone does not establish its cause.