Do you trust MT5 backtester ?

I recently switched mainly to MT5 because of the superiority of its backtester compared to MT4

I find more restrictive and less forgiving, which I find good, because we all need a realistic backtester.
but is it too restrictive ? Meaning, it will make some EAs not working well on the backtester but working live very well ?

A guy made a video on YT a whileback comparing live data with backtest (using "every ticks" AND "every ticks based on history")

and he found out that both price types were a bit wrong, and to have the same profitability than live, you would need to mix both prices
the "every ticks based on history" was making the EA lose almost all the time
while the "every ticks" mode was too optimistic.
It was a scalper mind you, but the results are nonetheless relevant

let me know what you found out about MT5's backtesting engine

ionone wrote: Mon May 11, 2026 6:23 pm I recently switched mainly to MT5 because of the superiority of its backtester compared to MT4

Mostly yes, with two big conditions. I rebuilt a few EAs in python on real tick data and compared them trade by
trade with the MT5 tester, so I got to see exactly where they differ:
1. Tick model. "Every tick" generates ticks from M1 bars; "Every tick based on real ticks" uses the broker's
history. For EAs with tight SL/TP, trailing or anything that fires inside a bar, results can be very
different. Real ticks is the one to trust.
2. Spread. Check what spread the test actually used. On gold the spread changes a lot by hour and around news,
and a fixed spread hides that.
With rel ticks and the broker's own history, my python results and the tester matched almost trade for trade,
so I don't think it's "too restrictive". If an EA only works on MT4's tester, it was usually the tester being
too generous, not MT5 being too strict.