MT4 backtesting with tick data

#1

Hey all,

I downloaded tick data from freetickdata(.com) and freetradingdata(.com) which seems to be a decent source of direct FXT files that one can use within MT4 to backtest an EA with 99% modelling quality. Funny thing is, every time I try to do it (after following instructions to get the FXT in the right place), MT4 squeaks telling me there’s no tick data within the time interval I selected, since you need to select 1/1/2020 - 12/31/2020 if you’re testing on 2020 tick data. Any idea why this happens or how to get around it? I tried downloading other intervals where the website clearly gives the dates that the data is for, and same issue. Any help would be greatly appreciated!

#2

Just a quick comment. Forget about 99% modeling quality. It’s a meaningful number.

#3

Really? I saw in a number of places that EAs fed with >99% and <99% quality had vastly different results in backtesting on the same period in time with the same settings. That was a little surprising…

1 Like
#4

I gave up with MT4 backtesting and instead write backtesting scripts on python. It’s relatively easy and quick to learn, an extra benefit is that you learn to code on concrete examples which is always faster.

1 Like
#5

Yes, in many cases the higher the quality shown by MetaTrader, the better the simulation. My point is that this is just a meaningful number. It does not mean that the modelling is good enough. It is just the way MetaTrader calculates modelling quality.

If my memory doesn’t fail, when I digged into the formula, I found that it is only possible to have values between 20 and 90% quality based on this MetaTrader formula. When you see some values over 90%, probably it is because some third party software is modifying that result internally.