I do not have a specific question, it's more of a general & conceptual one.
What would be the optimal approach to finding support and resistance levels?
Have you approached this problem before and what resources have you found useful? Here are some I've looked into:
https://medium.com/@judopro/using-machine-learning-to-programmatically-determine-stock-support-and-resistance-levels-9bb70777cf8e (a rather interesting starting point, but simple)
https://github.com/wilsonfreitas/awesome-quant#python (Looked through this, did not find anything that will identify support & resistance levels... did I miss something? Lots of other great things in there though)
The aim: to find the most important support and resistance levels, maybe in 3 categories, long, mid and short term. It will both be interesting to see what levels the alg would choose but mainly, it would be very interesting to use these levels in other more sophisticated deep learning algorithms, maybe combine it with a LSTM prediction to predict if the identified level will hold or break? Or maybe something else, Q-learning where the aim is to predict if it will hold or not?
Is unsupervised classification like K-Means needed, or should another type of alg be used? Have you found any great GitHub repos that you want to share? Or any frameworks? Or research?
What is the optimal way of identifying S & R and what would, in your opinion, be the best use of that data when it's extracted?
Thanks and hope you see where I am going with this!