[Tony Williamson]

My brother and I have been speaking about this and the best way to go about utilizing AI/ Machine learning in our expert advisor.

He has the opinion that while its very possible to train an AI to trade profitably, to stay ahead of the game you need huge amounts of computing power that only the top players in the industry have access to. We’re going to give it a go anyway 😅.

He also has experience in natural language processing and feels creating an AI to scrape data from reddit, twitter and other social sites could provide some great data to build a trading strategy from. I wouldn’t know how to even start building that program so I’ll leave that one to him.

[Bobby Belfort]

The future of AI is very bright. When applied to trading it might not be perfect at the moment because of some factors to consider such as the price, how AI can react to sentiments, news, manipulations,psychology, etc Nevertheless, its something people should look into. Still a preference game though.

[Samuel Junghenn]

I have been down the AI development route before multiple times. In the end it”s very expensive for the outcome at the end of the day.

With less time and money you can get the same result or better for 1/10th the cost and time with rule based systems that can be designed and tested very quickly.

If we ran a webinar about how to build and test bots quickly would you guys be interested?

[Tony Williamson]

I’d certainly be happy to join a webinar on the subject. I’ve attempted the rule based bots a few times but I feel I’m missing some vital step, would be interesting to see it from another angle.

[Daniel Lee]

I definitely see the potential in using AI / Machine Learning in trading.

I mean, identification and prediction are the name of the game for both trading and AI.

However, as no-one wants to give away their golden egg, especially not in the trading industry, I think it will be much more difficult for AI / Machine Learning to become widely used as a market predictor due to the technical aspects of building a capable model.

[Samuel Junghenn]

Ok, if we”ll add it to the list and when we get enough interest from others we”ll run it.

[stefan zarembski]

AI is certainly very good, insofar as we can enter the right economic indicators and sufficiently quickly. What do you suggest to obtain the right information certified sincere, because we have just seen that there is a lot of manipulation at the highest level, at the level of currencies and raw materials, which can suddenly reverse trends without any real financial logic. We can only follow the direction of the major trends.

[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.