The MetaTrader Strategy Tester: How to Use It
The strategy tester produces numbers quickly, and that is its main danger: the settings that affect the result more than the strategy itself all sit in one window and are easy to skip past.
The modelling mode matters more than the strategy
The tester offers several modes: open prices only, control points, and every tick. For a strategy deciding once a day on the close, the difference is small. For everything else it is fundamental: the open-prices mode assumes nothing happened inside the bar, and stops and targets within a single bar are handled approximately. Testing intrabar logic in that mode is pointless.
History quality
The modelling quality figure is the first thing to look at in a report. A low value means the tester filled in missing data, and the result applies to the reconstructed history rather than the real one. This matters especially on lower timeframes and older periods, where real ticks may not exist in the archive at all.
Spread and costs
By default the tester often uses the current or a fixed spread, whereas the real one varied through the day and widened on news. For strategies with targets of tens of pips this is secondary; for scalping it determines the sign of the result. Set the spread manually and with a margin rather than trusting the value supplied.
Choosing a period: two common errors
The first is testing on a stretch you already examined during development; that is recollection, not verification. The second is choosing a period that falls entirely within one market regime. A sensible minimum includes both trending and ranging sections, and the final check is done on a stretch that took no part in the tuning.
This material is educational and is not individual investment advice. Trading forex carries the risk of losing capital.
Frequently asked questions
Which modelling mode should I choose in the tester?
For strategies deciding on the daily close, control points suffice. Any intrabar logic requires every tick — in simplified modes the behaviour of stops and targets inside a bar is approximated.
What does modelling quality mean in the report?
The share of data the tester did not reconstruct itself. A low value means part of the history is synthetic and the result applies to that rather than to real quotes.
What spread should I set for testing?
Set it manually with a margin over the average rather than trusting the supplied value. Real spreads vary through the day and widen around releases.
New backtest studies — to your inbox
When a new reproducible study comes out or the data is updated, we send a short email. No spam, no "signals", no selling your address.
By leaving your email you agree to receive occasional emails from the portal. Unsubscribe in one click from any of them.