With machine learning being on a rise, I wanted to know if anyone has ever tried any machine learning strategy to trade?
I tried an auto Optimization based on machine learning and it worked pretty well, especially when considering how much time I saved. This one. I will get back to Optimizing after I trial the system for another month, I’ve been trading it for about 3 weeks until now.
Hey, I am also interested in this. Have you been able to find anything that works even if it is not 100% accurate
Hi there,
I’m working on this.
I’ve used Facebook’s ‘Prophet’ tool which is time series machine learning. - Results seem mixed.
I have built ARIMA, LSTM and random forest models too, which I’m currently training and testing.
I’ll gladly share an overview of the results.
It’s all written in Python and using a library called ‘TA’ for indicators along with a few others that I’ve coded up.
If anyone else is working on this and would like to compare notes reach out!
Hey @rllewellyn,
Did you make any progress on this? I’m interested in starting a machine learning project for that
is it worth it or not?
I am still working on it. I’m using NEAT and you can follow my progress here: Making an Ai trader
Whether it’s worth or not - I’m not sure.
I’ve been mucking around with it for a while around job and other projects but I’ve not got a profitable bot that I can deploy yet.
thank you for sharing the experience with us,
I’m wondering if your model can detect supports and resistance zones,
and what about the trend line …? I’m starting with AI NEAT Youtube Playlist
and I’m thinking we should start with defining Technical Analysis for the model,
like how he should know the trend and the support and resistance zones …
Currently my bot takes a few EMA’s MACD, RSI, StockRSI and Awesome Oscillator as inputs as well as OCHL for the last two candles. currently I don’t have a support / resistance or trend line input.
I use TA lib for python which gives a lot of normal indicators out of the box
good, I believe if we can put conditions for the bot like “don’t go long when downtrend …” it will make a difference, so that’s why I started thinking about the Technical Analysis part, and how the bot should know his position before making decisions, for indicators is to double confirm this decision