You can also use your MT5 data in the Express Generator to create and validate your strategies.
Export data from MT with the “Data Export.mq5”. (You can download the script from EA Studios, Data Import page.) Set the proper count of bars and commission values in the script's Input.
Place the files in a subdirectory of ExGen's data directory. For example: C:\express-generator\data\MyData
Use the directory's name as a server name when running the generator.
node .\bin\gen.js --server MyData --symbol EURUSD --period M15 --max-data-bars 100000
This is exactly what I've been doing, since I bought EG. In the past few days I've been using Darwinex data instead of my own because I thought it would be the same as the Darwinex demo account I'm testing on, so I thought I could skip that step. Turns out I might not be able to and get consistent results.
]]>Export data from MT with the “Data Export.mq5”. (You can download the script from EA Studios, Data Import page.) Set the proper count of bars and commission values in the script's Input.
Place the files in a subdirectory of ExGen's data directory. For example: C:\express-generator\data\MyData
Use the directory's name as a server name when running the generator.
node .\bin\gen.js --server MyData --symbol EURUSD --period M15 --max-data-bars 100000> I uploaded MT5 tick data from my Darwinex-Demo account to EA Studio
Are the MT and the EA Studio backtest results similar now?
A lot closer for sure. I don't expect real ticks to match open bars, but EA Studio was a lot closer to MT5 1min OHLC with banded indicators. I'm generating new strategies overnight with express generator on exported history from MT5 and I'll do more comparing in EA Studio tomorrow, particularly with Bollinger Bands.
]]>Are the MT and the EA Studio backtest results similar now?
We use Darwinex data for EA Studio and Express Generator from: Darwinex-Live through Access Server LIVE MT5 New
]]>Here is an example strategy that tests very well on EA Studio and Express Generator for the past year on Darwinex (in fact every broker in the drop down list), but does not give the same results when backtested in MT5 on Darwinex demo account. I'm excluding 22:00 to 02:00 for rollover to not get any hours that would have an average spread lower than reality during the beginning or end of a bar. I'm using "every tick based on real ticks" in MT5, but even 1 minute OHLC gives similar to real ticks for backtest results in MT5 compared to EA Studio website. This strategy has bollinger bands as an entry condition. Try downloading the strategy and comparing EA Studio backtest to MT5 backtest. How do we solve this, other than just disabling bollinger bands and similar banded indicators when generating strategies?
I'm getting somewhere with this. I uploaded MT5 tick data from my Darwinex-Demo account to EA Studio and it's significantly different than the Darwinex server data, enough that some strategies and indicators produce 12 losing trades instead of 184 trades with a PF>1.5 for example. Quite a difference.
]]>This happens with the "Bands" indicators, such as Bollinger Bands, Envelopes, and Donchian Channels.
A possible solution, in addition to Max Spread Protection, is to disable trading when gaps are possible during the Trading session like:
- Session open: 01:00
- Session close: 23:00
- Trade on Sunday - off
I'm already doing that in express generator though. I ended up just disabling those indicators for entry and exit.
]]>A possible solution, in addition to Max Spread Protection, is to disable trading when gaps are possible during the Trading session like:
- Session open: 01:00
- Session close: 23:00
- Trade on Sunday - off
The real tick backtest improves, as you forecasted, BUT, the trades are very few and the profit is just +100$ in 2025 vs around +900$ in Open Price.
Now, the big question: is Eastudio backtest more accurate than MT5 Real Tick backtest?
If not, the spread EAs studio imports with the Data Export script from MT5 is the Real Ticks MT5 spread, or the Averaged Bar MT5 Spread you mentioned above?
Set Max Spread between 30 and 50 points. It will eliminate the overestimated backtest results and may even make previously losing strategies profitable.
]]>But, the backtest was not only on Sunday night or Monday morning, but every day of the week. So it seems that only partially the negative real tick backtest is due to the gap, given that you say "usually on Sunday evening or on Monday morning":
]]>This is problematic on bar opening after a gap (usually on Sunday evening or on Monday morning). The spread can be high for a few seconds or minutes, but then lower later. It results in a low average spread for the bar. However, when we trade, we open a position at the Bar open with a high spread, which may ruin the position.
I have an idea to catch such cases by artificially increasing the spreads of bars after a gap.
I'll gather information on real spreads and gaps, and see what the RealSpread/AverageSpread ratio is for the different symbols and time frames.
The EA Studio backtest works perfectly with the data we feed it.
I'm also considering an advanced data feed with two spreads per bar—one at bar open and one averaged.
...
The spread you set in the Symbol Settings is valid for market data without a real spread. (Data you may import from MetaTrader 4.)
]]>I had viewn 20 points from Custom Settings, so not clear what it is 20 points at this point, if in fact I see from the chart that real spreads also on EaStudio jumps from 20 points to 140 points 
.
To recap: at the beginning you meant that, even with this so high real Spreads on EAstudio, some indicators (like Bollinger Bands) change those 140 points spreads into very low ones?
Hope to read soon good news of having this fixed ;-) Bollinger Bands is one of my preferred indicators.
You're saying that this is the only reason why there's mismatches among EaStudio backtest results and Mt5? Please it would be good to have a complete list of these indicators that have this issue, so that we can avoid to use them while they get fixed as you say you're working on that.
]]>You can check that in the Indicator Chart.
