A backtest is an experiment under chosen assumptions. It can reveal logic, trade frequency and sensitivity, but it is not a forecast and cannot reproduce every live quote, spread or execution event.
Configure MT4 Strategy Tester
- Open View → Strategy Tester or press Ctrl+R.
- Select the Expert Advisor, symbol and vendor-recommended period.
- Choose a meaningful date range that includes different conditions.
- Set model, spread, deposit and EA inputs; record them.
- Run a short visual test first to catch obvious logic or data problems.
Understand data and modeling limits
Historical-data quality, tick construction, time zone, spread and commission change the result. “Every tick” in legacy MT4 is still derived from the available history. Fixed spread can understate stress around rollover or news. If the EA depends on intrabar order, the modeling choice matters more.
Do not compare two reports unless symbol, dates, model, spread, deposit and inputs are comparable.
Read the report beyond net profit
| Metric | Question it answers | Warning sign |
|---|---|---|
| Maximal / relative drawdown | How deep was the historical decline? | Risk is unacceptable even if final profit is high. |
| Number of trades | Is the sample large enough to interpret? | A few trades dominate the conclusion. |
| Profit factor | Gross profit relative to gross loss? | Small assumption changes erase the edge. |
| Average trade | Can spread and slippage consume the edge? | Average outcome is tiny versus transaction costs. |
| Concentration | Does one period create most profit? | Long stagnation or one exceptional cluster. |
Test robustness, not the prettiest curve
Move the date range, increase spread, vary inputs modestly and test an untouched out-of-sample period. A stable system should not collapse because one parameter changes by a single point. Avoid choosing the best combination from thousands of trials without a separate validation period.
Continue with forward demo observation
Demo testing reveals current broker symbols, trading hours, connection behavior and log errors. Compare forward trade timing with the expected logic. Live execution can still differ, so use small defined risk if you later proceed.
Choose a model that matches the EA’s decisions
| Standard MT4 mode | What it can establish | Important limit |
|---|---|---|
| Open prices only | A quick check of rules that act on completed bars. | Unsuitable when entry or position management depends on movement inside a bar. |
| Control points | A rough preliminary estimate. | Not a reliable final validation of intrabar behavior. |
| Every tick | A more detailed simulated price path using available lower-timeframe history. | Standard MT4 ticks are modeled; this is not automatically a replay of recorded real ticks. |
The standard tester stores Bid history and constructs Ask using its selected spread. The default can use the spread at test start; a custom spread remains a simplifying assumption. Variable live spreads, commissions, swaps and execution effects need separate verification. Do not describe a fixed-spread run as a full reconstruction of live trading.
Keep a reproducible test record
Save the EA version, Inputs, exact symbol, contract properties, date range, model, spread, deposit, account currency and source of history with the report. Note missing data and whether the initial indicator warm-up affects the first trades. Re-run an unchanged configuration to check that the baseline is reproducible.
Review open exposure and the treatment of positions at the end of the test, not just closed-trade profit. Inspect the actual report and chart legend; do not infer equity from a balance-only curve. Keep a later chronological period untouched until the rules are fixed, then observe a forward demo. If you tune after seeing that period, it is no longer an independent test.
The official MT4 tester settings describe the modeling modes and spread assumptions.
Official reference
MetaTrader documents Strategy Tester configuration and testing journals in its official testing help.