[Nuno Miguel Madeira Amaro Pires Costa]

Just my opinion / Experience:

I am new here on TwE but I am firmly in the prop firm scene for about 4 years now.

I coded and maintained a MT4 EA during ’19 and ’20, which then turned into MT5.

With under 2000 lines of MQL code I have 7 entry types, with simple boolean logic, supported by 3-4 indicators. No AI or NLP bells and whistles fanciness.

MT5 platform lets you backtest your code and real world results are not distant from the back tested if you use a meaningful sample.

I find it enough in terms of logic.

It could be better – YES!

But it is sufficient and easy to get to.

I do not know how far into the AI/NLP trading bot path are you but I would definitely suggest making a small easy script and improve on that, while you run it on a demo account or low fund real account, instead of investing 12 to 24 months making something theoretically near perfect but that may not perform anywhere near what you expect it to, due to one of the many possible bottlenecks.

If you can do reliably 2% a month you will have MANY prop firms throw money at you. Real money. If you aim for 10% a month you will find that your bot (you) will fail to be reliable on delivering that expectation.

Ride the turtle reality & ignore the hare dreams.