sands Thank you for your input! I may look into using it on bigger time frames for general market context. But I decided to put it on the side for now.
NJAMC I actually got the chance to speak to a machine learning expert working at MATLAB and he gave me a similar answer to yours. He mentioned that ARIMA+GARCH may not be appropriate for for smaller time frames on intraday data because of higher noise as you mentioned. He instead recommended I look into the Heston model which I've yet to check out. I'm also looking to using ML for market prediction. But my ultimate goal is to use ML and train it on data to find profitable strategies and validate it (which I understand will be a very complex process, if even possible). Thank you for your input as well! Been following your threads, keep up the good work