MT5 backtests: real ticks versus generated ticks

The Strategy Tester modes

Open prices only is the fastest and the least realistic: the EA sees one price per bar. Every tick generates a path inside each bar from M1 data. Every tick based on real ticks replays the actual tick stream of the broker. For anything that reacts inside a bar, only the last one is a serious test.

Why generated ticks flatter strategies

Modelled ticks follow a neat path inside each bar. Real markets gap, spike and widen the spread. Grids, scalpers and trailing logic look safer on the clean path than they are on the real one.

History quality and tick coverage

After a test, the report shows the history quality. A low value means the tester filled gaps. Prefer runs where the real tick history covers the whole period, and be careful with the first years of any broker archive.

Spread, commission and swap

Use the broker's real spread behaviour, a commission that matches the account type, and the actual swap. Gold and index CFDs change their costs through time, so test on conditions close to today's.

Execution delay

Set a delay such as 500 ms for systems that open and close often. If the profit disappears with a small delay, the edge was a simulation artefact.

Read the report, not the headline

Check number of trades, profit factor, maximum equity drawdown, the longest losing streak and the monthly distribution. Then reconcile with the tester log: symbol, timeframe, dates and EA name must be the ones you wanted.

How Eudora is tested

Every published number comes from the MT5 Strategy Tester on real broker ticks. Albireo adds a 500 ms delay on every order and still shows 69 of 69 months positive. Crux passed 20 of 20 stress scenarios, with spreads, delays up to 3 seconds and different start dates. When a result survives that, it is a result.