I'll remove the "Start from bar 100" option from EA Studio.

Now the "Start date" of the MT tester sets from what day the tester starts opening positions.

When we set "start date", it means the start of the actual deals. The actual data set start is unclear. Probably MT uses the complete data set.

I'm going to add a ne woption in the "Data Horizon":  Trade start

We will be able to set individually three params:
- Data start
- Trade start
- Data end

You are right!

The first trade is on 2018.07.03 at 02:30.

This is strange because before the trading started at bar 101. We even have such an option in EA Studio.
However, it looks like they load one month before the set date.

What to do?
Do I need to add the option "Load one-month preliminary data"?

I'm not sure what the right thing is to do.  Moving Average-based indicators need some preliminary data in order to begin drawing their shape.  You probably do not want to allow trades before all the indicators are "warmed-up".

I don't think that it is necessary to add an option.  Rather, I would **always** preload 100? 500? bars and allow the calculations to begin further back in history, but ignore all trade signals until the 'From' date is reached.

When you manually / discretionary trade you always have a chart full of bars and pre-drawn indicator markings to view before you make a decision to trade.  Without that preliminary data a discretionary trader would need to wait until there were enough bars for their indicators to have begun working and displaying their patterns.

The first trade is on 2018.07.03 at 02:30.

This is strange because before the trading started at bar 101. We even have such an option in EA Studio.
However, it looks like they load one month before the set date.

What to do?
Do I need to add the option "Load one-month preliminary data"?

How can I make a reliable comparison of the EA Studio and MT performance when I cannot set the backtest start date?

If the chart shows a later date, OK. I'll assume there is no data.
However, what if it shows an earlier date?

I want a perfect backtest match to address users' questions in advance. Now I'm working on a new Generator engine and found a way to make some things better. For example - closing at a bar's Open when a Trailing Stop appears in the gap between the previous Close and the current Open.

Hello Popov,

MT4's Strategy Tester can be so unreliable and no one knows how it works behind the scenes.  I've seen data gaps of greater than 2 weeks and Strategy Tester chugs along with no warning.  I'm not sure it should be considered the "gold standard" that other back-testing software should strive to mimic.

With regards to the earlier date on the chart -- perhaps Strategy Tester "front-loads" some extra bars to accommodate certain types of indicators, but only begins Entering / Exiting trades based on the 'From' date that you entered.   In the event you see an earlier date on the chart, under the 'Results' tab do you also see trades prior to your 'From' date?  I never have -- though I don't always look that carefully.

If the chart shows a later date, OK. I'll assume there is no data.
However, what if it shows an earlier date?

I want a perfect backtest match to address users' questions in advance. Now I'm working on a new Generator engine and found a way to make some things better. For example - closing at a bar's Open when a Trailing Stop appears in the gap between the previous Close and the current Open.